vecId stringlengths 12 23 | id stringlengths 2 13 | conference stringclasses 11
values | year float64 2.02k 2.03k | title stringlengths 6 189 | abstract stringlengths 10 4.74k | author stringlengths 0 7.45k | aff stringlengths 0 7.16k | status stringclasses 11
values | track stringclasses 4
values | keywords stringlengths 0 804 | github stringlengths 0 141 | site stringlengths 0 193 | gsCitation float64 -1 11.1k | arxiv stringlengths 0 12 | text stringlengths 58 4.82k | vector list |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
corl_2024_0gDbaEtVrd | 0gDbaEtVrd | corl | 2,024 | One Model to Drift Them All: Physics-Informed Conditional Diffusion Model for Driving at the Limits | Enabling autonomous vehicles to reliably operate at the limits of handling— where tire forces are saturated — would improve their safety, particularly in scenarios like emergency obstacle avoidance or adverse weather conditions.
However, unlocking this capability is challenging due to the task's dynamic nature and the ... | Franck Djeumou;Thomas Jonathan Lew;NAN DING;Michael Thompson;Makoto Suminaka;Marcus Greiff;John Subosits | ;Toyota Research Institute;Toyota Research Institute;Toyota Research Institute;;Mitsubishi Electric Research Labs;Toyota Motor Corporation | Poster | main | Diffusion Models;Learning for Control;Autonomous Drifting;Model Predictive Control | https://openreview.net/forum?id=0gDbaEtVrd | 8 | One Model to Drift Them All: Physics-Informed Conditional Diffusion Model for Driving at the Limits
Enabling autonomous vehicles to reliably operate at the limits of handling— where tire forces are saturated — would improve their safety, particularly in scenarios like emergency obstacle avoidance or adverse weather con... | [
-0.08241964876651764,
-0.032408084720373154,
-0.0076278685592114925,
-0.008603793568909168,
-0.056566838175058365,
-0.014160123653709888,
-0.012650201097130775,
0.038742393255233765,
0.013929951936006546,
0.025631848722696304,
-0.03769281134009361,
0.015881802886724472,
0.014012813568115234,... | ||
corl_2024_1IzW0aniyg | 1IzW0aniyg | corl | 2,024 | EscIRL: Evolving Self-Contrastive IRL for Trajectory Prediction in Autonomous Driving | While deep neural networks (DNN) and inverse reinforcement learning (IRL) have both been commonly used in autonomous driving to predict trajectories through learning from expert demonstrations, DNN-based methods suffer from data-scarcity, while IRL-based approaches often struggle with generalizability, making both hard... | Siyue Wang;Zhaorun Chen;Zhuokai Zhao;Chaoli Mao;Yiyang Zhou;Jiayu He;Albert Sibo Hu | CIDI Intelligent Driving(Chengdu) Technology Co., Ltd.;University of Chicago;University of Chicago;CIDI;Xi'an Jiaotong University;cidi-lab;CiDi | Poster | main | Reinforcement Learning;Trajectory Prediction;Autonomous Driving | https://github.com/SiyueWang-CiDi/EscIRL | https://openreview.net/forum?id=1IzW0aniyg | 2 | EscIRL: Evolving Self-Contrastive IRL for Trajectory Prediction in Autonomous Driving
While deep neural networks (DNN) and inverse reinforcement learning (IRL) have both been commonly used in autonomous driving to predict trajectories through learning from expert demonstrations, DNN-based methods suffer from data-scarc... | [
-0.10358572751283646,
-0.02554372139275074,
0.03456570580601692,
-0.00735588651150465,
-0.022332191467285156,
-0.00653908122330904,
-0.01786760985851288,
0.002717269118875265,
0.00391231058165431,
0.025933559983968735,
-0.022406447678804398,
-0.013597947545349598,
-0.05583976209163666,
0.0... | |
corl_2024_1TEZ1hiY5m | 1TEZ1hiY5m | corl | 2,024 | Learning Robotic Locomotion Affordances and Photorealistic Simulators from Human-Captured Data | Learning reliable affordance models which satisfy human preferences is often hindered by a lack of high-quality training data. Similarly, learning visuomotor policies in simulation can be challenging due to the high cost of photo-realistic rendering. We present PAWS: a comprehensive robot learning framework that uses a... | Alejandro Escontrela;Justin Kerr;Kyle Stachowicz;Pieter Abbeel | University of California, Berkeley;University of California, Berkeley;University of California, Berkeley;Covariant | Poster | main | Navigation;Dataset;Real2Sim | https://openreview.net/forum?id=1TEZ1hiY5m | 1 | Learning Robotic Locomotion Affordances and Photorealistic Simulators from Human-Captured Data
Learning reliable affordance models which satisfy human preferences is often hindered by a lack of high-quality training data. Similarly, learning visuomotor policies in simulation can be challenging due to the high cost of p... | [
-0.07279437780380249,
-0.03138977289199829,
-0.02232784405350685,
0.017638063058257103,
-0.03934932127594948,
-0.006511509418487549,
0.03912510722875595,
0.014443032443523407,
-0.003031542059034109,
0.05956583097577095,
-0.01861899346113205,
0.013807762414216995,
0.010808918625116348,
0.02... | ||
corl_2024_1jc2zA5Z6J | 1jc2zA5Z6J | corl | 2,024 | Get a Grip: Multi-Finger Grasp Evaluation at Scale Enables Robust Sim-to-Real Transfer | This work explores conditions under which multi-finger grasping algorithms can attain robust sim-to-real transfer. While numerous large datasets facilitate learning *generative* models for multi-finger grasping at scale, reliable real-world dexterous grasping remains challenging, with most methods degrading when deploy... | Tyler Ga Wei Lum;Albert H. Li;Preston Culbertson;Krishnan Srinivasan;Aaron Ames;Mac Schwager;Jeannette Bohg | Stanford University;;California Institute of Technology;Stanford University;California Institute of Technology;Stanford University;Stanford University | Poster | main | Multi-Fingered Grasping;Large-Scale Grasp Dataset;Sim-to-Real | https://github.com/tylerlum/get_a_grip | https://openreview.net/forum?id=1jc2zA5Z6J | 2 | Get a Grip: Multi-Finger Grasp Evaluation at Scale Enables Robust Sim-to-Real Transfer
This work explores conditions under which multi-finger grasping algorithms can attain robust sim-to-real transfer. While numerous large datasets facilitate learning *generative* models for multi-finger grasping at scale, reliable rea... | [
-0.03876669332385063,
-0.003993328660726547,
-0.028224874287843704,
-0.027979275211691856,
-0.03162546083331108,
-0.01708795130252838,
-0.016218911856412888,
0.02588224783539772,
0.012364912778139114,
0.01879769004881382,
-0.02739361859858036,
0.03145543113350868,
-0.012100422754883766,
0.... | |
corl_2024_1tCteNSbFH | 1tCteNSbFH | corl | 2,024 | Trajectory Improvement and Reward Learning from Comparative Language Feedback | Learning from human feedback has gained traction in fields like robotics and natural language processing in recent years. While prior works mostly rely on human feedback in the form of comparisons, language is a preferable modality that provides more informative insights into user preferences. In this work, we aim to i... | Zhaojing Yang;Miru Jun;Jeremy Tien;Stuart Russell;Anca Dragan;Erdem Biyik | University of Southern California;University of Southern California;University of California, Berkeley;University of California, Berkeley;University of California, Berkeley;University of Southern California | Poster | main | Learning from Human Language Feedback;Reward Learning;Human-Robot Interaction | https://github.com/USC-Lira/language-preference-learning | https://openreview.net/forum?id=1tCteNSbFH | 8 | Trajectory Improvement and Reward Learning from Comparative Language Feedback
Learning from human feedback has gained traction in fields like robotics and natural language processing in recent years. While prior works mostly rely on human feedback in the form of comparisons, language is a preferable modality that provi... | [
-0.07589001953601837,
-0.026844175532460213,
0.0035684595350176096,
-0.01956217736005783,
-0.017877396196126938,
0.01708180643618107,
0.01022100355476141,
-0.015986697748303413,
0.005335139576345682,
0.004511468578130007,
-0.023455893620848656,
-0.049532558768987656,
-0.041557926684617996,
... | |
corl_2024_2AZfKk9tRI | 2AZfKk9tRI | corl | 2,024 | Multi-agent Reinforcement Learning with Hybrid Action Space for Free Gait Motion Planning of Hexapod Robots | Legged robots are able to overcome challenging terrains through diverse gaits formed by contact sequences. However, environments characterized by discrete footholds present significant challenges. In this paper, we tackle the problem of free gait motion planning for hexapod robots walking in randomly generated plum blo... | Huiqiao Fu;Kaiqiang Tang;Peng Li;Guizhou Deng;Chunlin Chen | Nanjing University;Nanjing University;Institute of Software, Chinese Academy of Sciences;;Nanjing University | Poster | main | Free Gait;Hexapod Robot;Hybrid Action Space;Multi-agent Reinforcement Learning | https://openreview.net/forum?id=2AZfKk9tRI | 0 | Multi-agent Reinforcement Learning with Hybrid Action Space for Free Gait Motion Planning of Hexapod Robots
Legged robots are able to overcome challenging terrains through diverse gaits formed by contact sequences. However, environments characterized by discrete footholds present significant challenges. In this paper, ... | [
-0.06565209478139877,
-0.04211918264627457,
-0.027843035757541656,
0.016576699912548065,
-0.008565688505768776,
-0.013175884261727333,
0.01770424284040928,
-0.0015992468688637018,
0.023060070350766182,
0.014030634425580502,
-0.02924337051808834,
-0.018358943983912468,
-0.01213927287608385,
... | ||
corl_2024_2CScZqkUPZ | 2CScZqkUPZ | corl | 2,024 | Genetic Algorithm for Curriculum Design in Multi-Agent Reinforcement Learning | As the deployment of autonomous agents in real-world scenarios grows, so does the interest in their application to competitive environments with other robots. Self-play in Reinforcement Learning (RL) enables agents to develop competitive strategies. However, the complexity arising from multi-agent interactions and the ... | Yeeho Song;Jeff Schneider | Carnegie Mellon University;Carnegie Mellon University | Poster | main | Reinforcement Learning;Multiagent Learning;Curricular Learning | https://github.com/yeehos/GEnetic-Multiagent-Selfplay | https://openreview.net/forum?id=2CScZqkUPZ | 1 | Genetic Algorithm for Curriculum Design in Multi-Agent Reinforcement Learning
As the deployment of autonomous agents in real-world scenarios grows, so does the interest in their application to competitive environments with other robots. Self-play in Reinforcement Learning (RL) enables agents to develop competitive stra... | [
-0.062280625104904175,
-0.008072194643318653,
-0.02093593031167984,
-0.01941092126071453,
-0.023214031010866165,
-0.006650735158473253,
-0.013423847034573555,
0.016822170466184616,
0.014393451623618603,
0.01168232411146164,
-0.048197824507951736,
-0.017697637900710106,
-0.04812251403927803,
... | |
corl_2024_2LLu3gavF1 | 2LLu3gavF1 | corl | 2,024 | Robot See Robot Do: Imitating Articulated Object Manipulation with Monocular 4D Reconstruction | Humans can learn to manipulate new objects by simply watching others; providing robots with the ability to learn from such demonstrations would enable a natural interface specifying new behaviors. This work develops Robot See Robot Do (RSRD), a method for imitating articulated object manipulation from a single monocula... | Justin Kerr;Chung Min Kim;Mingxuan Wu;Brent Yi;Qianqian Wang;Ken Goldberg;Angjoo Kanazawa | University of California, Berkeley;University of California, Berkeley;Xi'an Jiaotong University;University of California, Berkeley;University of California, Berkeley;University of California, Berkeley;University of California, Berkeley | Poster | main | Feature Fields;Visual Imitation;Grasping;Articulated Objects | https://github.com/kerrj/rsrd | https://openreview.net/forum?id=2LLu3gavF1 | 15 | Robot See Robot Do: Imitating Articulated Object Manipulation with Monocular 4D Reconstruction
Humans can learn to manipulate new objects by simply watching others; providing robots with the ability to learn from such demonstrations would enable a natural interface specifying new behaviors. This work develops Robot See... | [
-0.03082842193543911,
-0.006290764082223177,
0.0055412063375115395,
-0.0056561692617833614,
-0.020251845940947533,
0.036438606679439545,
-0.003494868753477931,
-0.013547215610742569,
0.00957410130649805,
0.021888915449380875,
-0.03914253041148186,
-0.02190730907022953,
0.02856595441699028,
... | |
corl_2024_2SYFDG4WRA | 2SYFDG4WRA | corl | 2,024 | Manipulate-Anything: Automating Real-World Robots using Vision-Language Models | Large-scale endeavors like RT-1 and widespread community efforts such as Open-X-Embodiment have contributed to growing the scale of robot demonstration data. However, there is still an opportunity to improve the quality, quantity, and diversity of robot demonstration data. Although vision-language models have been show... | Jiafei Duan;Wentao Yuan;Wilbert Pumacay;Yi Ru Wang;Kiana Ehsani;Dieter Fox;Ranjay Krishna | NVIDIA;University of Washington, Seattle;Universidad Nacional de Ingeniería;University of Washington;Allen Institute for Artificial Intelligence;Department of Computer Science;University of Washington | Poster | main | Robot Learning; Multimodal Large Language Model; Data Generation; Imitation Learning; Behavior Cloning | https://github.com/Robot-MA/manipulate-anything/tree/main | https://openreview.net/forum?id=2SYFDG4WRA | 39 | Manipulate-Anything: Automating Real-World Robots using Vision-Language Models
Large-scale endeavors like RT-1 and widespread community efforts such as Open-X-Embodiment have contributed to growing the scale of robot demonstration data. However, there is still an opportunity to improve the quality, quantity, and divers... | [
-0.02514158934354782,
-0.008075951598584652,
0.00026492003235034645,
0.05216602981090546,
-0.018597763031721115,
0.007120681460946798,
0.0010187235893681645,
0.029239557683467865,
0.022501908242702484,
0.0035764926578849554,
-0.030531710013747215,
-0.016179591417312622,
0.011389398947358131,... | |
corl_2024_2sg4PY1W9d | 2sg4PY1W9d | corl | 2,024 | Learning Transparent Reward Models via Unsupervised Feature Selection | In complex real-world tasks such as robotic manipulation and autonomous driving, collecting expert demonstrations is often more straightforward than specifying precise learning objectives and task descriptions. Learning from expert data can be achieved through behavioral cloning or by learning a reward function, i.e., ... | Daulet Baimukashev;Gokhan Alcan;Kevin Sebastian Luck;Ville Kyrki | Aalto University;Aalto University;Vrije Universiteit Amsterdam;Aalto University | Poster | main | Inverse reinforcement learning;Reinforcement learning;Imitation learning;Robots;Reward learning;Robot learning | https://github.com/baimukashev/reward-learning | https://openreview.net/forum?id=2sg4PY1W9d | 0 | Learning Transparent Reward Models via Unsupervised Feature Selection
In complex real-world tasks such as robotic manipulation and autonomous driving, collecting expert demonstrations is often more straightforward than specifying precise learning objectives and task descriptions. Learning from expert data can be achiev... | [
-0.07196776568889618,
-0.015986589714884758,
-0.00032329271198250353,
0.008860094472765923,
-0.002367299050092697,
-0.027737576514482498,
0.006587674375623465,
0.020503317937254906,
0.009253667667508125,
0.016895558685064316,
-0.026144539937376976,
-0.011020064353942871,
-0.02031590044498443... | |
corl_2024_3NI5SxsJqf | 3NI5SxsJqf | corl | 2,024 | Accelerating Visual Sparse-Reward Learning with Latent Nearest-Demonstration-Guided Explorations | Recent progress in deep reinforcement learning (RL) and computer vision enables artificial agents to solve complex tasks, including locomotion, manipulation, and video games from high-dimensional pixel observations. However, RL usually relies on domain-specific reward functions for sufficient learning signals, requirin... | Ruihan Zhao;ufuk topcu;Sandeep P. Chinchali;Mariano Phielipp | University of Texas at Austin;University of Texas, Austin;Intel Labs;University of Texas at Austin | Poster | main | Computer Vision;Sparse Reward;RL from Demonstrations | https://openreview.net/forum?id=3NI5SxsJqf | 0 | Accelerating Visual Sparse-Reward Learning with Latent Nearest-Demonstration-Guided Explorations
Recent progress in deep reinforcement learning (RL) and computer vision enables artificial agents to solve complex tasks, including locomotion, manipulation, and video games from high-dimensional pixel observations. However... | [
-0.0834430605173111,
0.0011691030813381076,
0.000011807649570982903,
0.019890496507287025,
-0.040191490203142166,
0.024238044396042824,
0.04127371311187744,
0.014488715678453445,
0.0060081989504396915,
0.052879612892866135,
-0.026122603565454483,
-0.012697450816631317,
-0.021140649914741516,... | ||
corl_2024_3ZAgXBRvla | 3ZAgXBRvla | corl | 2,024 | FlowBotHD: History-Aware Diffuser Handling Ambiguities in Articulated Objects Manipulation | We introduce a novel approach to manipulate articulated objects with ambiguities, such as opening a door, in which multi-modality and occlusions create ambiguities about the opening side and direction. Multi-modality occurs when the method to open a fully closed door (push, pull, slide) is uncertain, or the side from w... | Yishu Li;Wen Hui Leng;Yiming Fang;Ben Eisner;David Held | Tsinghua University;Carnegie Mellon University;Carnegie Mellon University;Carnegie Mellon University;Carnegie Mellon University | Poster | main | Ambiguity;Multi-modality;Occlusion;Articulated Objects;Diffusion | https://openreview.net/forum?id=3ZAgXBRvla | 0 | FlowBotHD: History-Aware Diffuser Handling Ambiguities in Articulated Objects Manipulation
We introduce a novel approach to manipulate articulated objects with ambiguities, such as opening a door, in which multi-modality and occlusions create ambiguities about the opening side and direction. Multi-modality occurs when ... | [
-0.026064615696668625,
-0.057717181742191315,
-0.03862813487648964,
0.04605373740196228,
-0.02265183813869953,
-0.02737722173333168,
-0.018610885366797447,
0.016004424542188644,
0.03281516209244728,
0.024864517152309418,
-0.03626544401049614,
-0.010491476394236088,
0.004495677072554827,
0.... | ||
corl_2024_3bcujpPikC | 3bcujpPikC | corl | 2,024 | FREA: Feasibility-Guided Generation of Safety-Critical Scenarios with Reasonable Adversariality | Generating safety-critical scenarios, which are essential yet difficult to collect at scale, offers an effective method to evaluate the robustness of autonomous vehicles (AVs). Existing methods focus on optimizing adversariality while preserving the naturalness of scenarios, aiming to achieve a balance through data-dri... | Keyu Chen;Yuheng Lei;Hao Cheng;Haoran Wu;Wenchao Sun;Sifa Zheng | Tsinghua University;The University of Hong Kong;Tsinghua University;Tsinghua University;Tsinghua University;Tsinghua University | Poster | main | Feasibility;Scenario Generation;Autonomous Driving | https://github.com/CurryChen77/FREA | https://openreview.net/forum?id=3bcujpPikC | 3 | FREA: Feasibility-Guided Generation of Safety-Critical Scenarios with Reasonable Adversariality
Generating safety-critical scenarios, which are essential yet difficult to collect at scale, offers an effective method to evaluate the robustness of autonomous vehicles (AVs). Existing methods focus on optimizing adversaria... | [
-0.05088040232658386,
0.0014859603252261877,
0.009876783937215805,
0.00214184052310884,
-0.025515025481581688,
0.026768330484628677,
0.0055697206407785416,
0.030098002403974533,
0.028882110491394997,
0.03786100447177887,
-0.019660038873553276,
0.00851124431937933,
0.012776218354701996,
0.0... | |
corl_2024_3i7j8ZPnbm | 3i7j8ZPnbm | corl | 2,024 | UMI-on-Legs: Making Manipulation Policies Mobile with Manipulation-Centric Whole-body Controllers | We introduce UMI-on-Legs, a new framework that combines real-world and simulation data for quadruped manipulation systems. We scale task-centric data collection in the real world using a handheld gripper (UMI), providing a cheap way to demonstrate task-relevant manipulation skills without a robot. Simultaneously, we s... | Huy Ha;Yihuai Gao;Zipeng Fu;Jie Tan;Shuran Song | Columbia University;Stanford University;Stanford University;Google;Stanford University | Poster | main | Manipulation;Visuo-motor Policy;Whole-body Controller | https://github.com/real-stanford/umi-on-legs | https://openreview.net/forum?id=3i7j8ZPnbm | 45 | UMI-on-Legs: Making Manipulation Policies Mobile with Manipulation-Centric Whole-body Controllers
We introduce UMI-on-Legs, a new framework that combines real-world and simulation data for quadruped manipulation systems. We scale task-centric data collection in the real world using a handheld gripper (UMI), providing ... | [
-0.0619102381169796,
-0.04926946386694908,
-0.051376257091760635,
0.015542242676019669,
-0.02860807441174984,
-0.015015543438494205,
-0.004687159787863493,
0.02539243921637535,
0.007161259185522795,
0.016346151009202003,
-0.04269034415483475,
0.014331758953630924,
0.04878896474838257,
0.01... | |
corl_2024_3jNEz3kUSl | 3jNEz3kUSl | corl | 2,024 | PointPatchRL - Masked Reconstruction Improves Reinforcement Learning on Point Clouds | Perceiving the environment via cameras is crucial for Reinforcement Learning (RL) in robotics. While images are a convenient form of representation, they often complicate extracting important geometric details, especially with varying geometries or deformable objects. In contrast, point clouds naturally represent this ... | Balazs Gyenes;Nikolai Franke;Philipp Becker;Gerhard Neumann | Karlsruher Institut für Technologie;Karlsruher Institut für Technologie;FZI Forschungszentrum Informatik ;Karlsruhe Institute of Technology | Poster | main | Point Clouds;Self-Supervised Learning;Reinforcement Learning | https://github.com/balazsgyenes/pprl | https://openreview.net/forum?id=3jNEz3kUSl | 0 | PointPatchRL - Masked Reconstruction Improves Reinforcement Learning on Point Clouds
Perceiving the environment via cameras is crucial for Reinforcement Learning (RL) in robotics. While images are a convenient form of representation, they often complicate extracting important geometric details, especially with varying ... | [
-0.0005533811636269093,
0.003310643369331956,
-0.002264579990878701,
-0.02245519310235977,
-0.01668255589902401,
0.006612210068851709,
0.018842756748199463,
0.02692081779241562,
-0.008500116877257824,
0.01292489655315876,
-0.044982995837926865,
-0.029879746958613396,
-0.00106081273406744,
... | |
corl_2024_3wBqoPfoeJ | 3wBqoPfoeJ | corl | 2,024 | Twisting Lids Off with Two Hands | Manipulating objects with two multi-fingered hands has been a long-standing challenge in robotics, due to the contact-rich nature of many manipulation tasks and the complexity inherent in coordinating a high-dimensional bimanual system. In this work, we share novel insights into physical modeling, real-time perception,... | Toru Lin;Zhao-Heng Yin;Haozhi Qi;Pieter Abbeel;Jitendra Malik | ;;University of California, Berkeley;Covariant;University of California, Berkeley | Poster | main | Bimanual Manipulation;Sim-to-Real;Reinforcement Learning | https://openreview.net/forum?id=3wBqoPfoeJ | 28 | Twisting Lids Off with Two Hands
Manipulating objects with two multi-fingered hands has been a long-standing challenge in robotics, due to the contact-rich nature of many manipulation tasks and the complexity inherent in coordinating a high-dimensional bimanual system. In this work, we share novel insights into physica... | [
-0.07024779915809631,
-0.0027922471053898335,
0.011767160147428513,
0.0257587730884552,
-0.03630154952406883,
-0.006841589696705341,
0.003927838988602161,
0.026057858020067215,
-0.013225204311311245,
0.004425536375492811,
-0.03054414689540863,
-0.005902273114770651,
-0.021253790706396103,
... | ||
corl_2024_46SluHKoE9 | 46SluHKoE9 | corl | 2,024 | Continuously Improving Mobile Manipulation with Autonomous Real-World RL | We present a fully autonomous real-world RL framework for mobile manipulation that can learn policies without extensive instrumentation or human supervision. This is enabled by 1) task-relevant autonomy, which guides exploration towards object interactions and prevents stagnation near goal states, 2) efficient policy l... | Russell Mendonca;Emmanuel Panov;Bernadette Bucher;Jiuguang Wang;Deepak Pathak | Carnegie Mellon University;;Boston Dynamics AI Institute;;Carnegie Mellon University | Poster | main | Continual Learning;Mobile Manipulation;Reinforcement Learning | https://openreview.net/forum?id=46SluHKoE9 | 3 | Continuously Improving Mobile Manipulation with Autonomous Real-World RL
We present a fully autonomous real-world RL framework for mobile manipulation that can learn policies without extensive instrumentation or human supervision. This is enabled by 1) task-relevant autonomy, which guides exploration towards object int... | [
-0.07026921212673187,
-0.01020097080618143,
0.0038463440723717213,
0.005133121274411678,
-0.023292528465390205,
-0.022677112370729446,
0.009576231241226196,
-0.011739508248865604,
0.01568375900387764,
0.025232017040252686,
-0.05314948409795761,
-0.02355361357331276,
0.011170716024935246,
0... | ||
corl_2024_4Of4UWyBXE | 4Of4UWyBXE | corl | 2,024 | RP1M: A Large-Scale Motion Dataset for Piano Playing with Bi-Manual Dexterous Robot Hands | Endowing robot hands with human-level dexterity is a long-lasting research objective. Bi-manual robot piano playing constitutes a task that combines challenges from dynamic tasks, such as generating fast while precise motions, with slower but contact-rich manipulation problems. Although reinforcement learning based app... | Yi Zhao;Le Chen;Jan Schneider;Quankai Gao;Juho Kannala;Bernhard Schölkopf;Joni Pajarinen;Dieter Büchler | Max Planck Institute for Intelligent Systems;;Max Planck Institute for Intelligent Systems;University of Southern California;Aalto University;;Aalto University;Max Planck Institute for Intelligent Systems, Max-Planck Institute | Poster | main | Bi-manual dexterous robot hands;dataset for robot piano playing;imitation learning;robot learning at scale | https://openreview.net/forum?id=4Of4UWyBXE | 2 | RP1M: A Large-Scale Motion Dataset for Piano Playing with Bi-Manual Dexterous Robot Hands
Endowing robot hands with human-level dexterity is a long-lasting research objective. Bi-manual robot piano playing constitutes a task that combines challenges from dynamic tasks, such as generating fast while precise motions, wit... | [
-0.06796083599328995,
-0.019620096310973167,
-0.05255991220474243,
0.03242897242307663,
-0.01860787346959114,
0.0005445427959784865,
0.02771787904202938,
0.026185356080532074,
-0.0025683974381536245,
0.034075018018484116,
-0.02713135816156864,
0.014388701878488064,
0.00401105172932148,
0.0... | ||
corl_2024_55tYfHvanf | 55tYfHvanf | corl | 2,024 | Bimanual Dexterity for Complex Tasks | To train generalist robot policies, machine learning methods often require a substantial amount of expert human teleoperation data. An ideal robot for humans collecting data is one that closely mimics them: bimanual arms and dexterous hands. However, creating such a bimanual teleoperation system with over 50 DoF is a s... | Kenneth Shaw;Yulong Li;Jiahui Yang;Mohan Kumar Srirama;Ray Liu;Haoyu Xiong;Russell Mendonca;Deepak Pathak | Carnegie Mellon University;Carnegie Mellon University;;Carnegie Mellon University;;Carnegie Mellon University;Carnegie Mellon University;Carnegie Mellon University | Poster | main | Dexterous Manipulation;Bimanual;Behavior Cloning | https://openreview.net/forum?id=55tYfHvanf | 21 | Bimanual Dexterity for Complex Tasks
To train generalist robot policies, machine learning methods often require a substantial amount of expert human teleoperation data. An ideal robot for humans collecting data is one that closely mimics them: bimanual arms and dexterous hands. However, creating such a bimanual teleope... | [
-0.023834049701690674,
-0.04077768698334694,
-0.038669146597385406,
-0.0038899767678231,
0.007083381991833448,
0.02219616435468197,
0.032249391078948975,
0.015512841753661633,
0.02033236436545849,
0.03689947724342346,
-0.026827426627278328,
-0.012284137308597565,
0.013140732422471046,
-0.0... | ||
corl_2024_56IzghzjfZ | 56IzghzjfZ | corl | 2,024 | IMAGINATION POLICY: Using Generative Point Cloud Models for Learning Manipulation Policies | Humans can imagine goal states during planning and perform actions to match those goals. In this work, we propose IMAGINATION POLICY, a novel multi-task key-frame policy network for solving high-precision pick and place tasks. Instead of learning actions directly, IMAGINATION POLICY generates point clouds to imagine de... | Haojie Huang;Karl Schmeckpeper;Dian Wang;Ondrej Biza;Yaoyao Qian;Haotian Liu;Mingxi Jia;Robert Platt;Robin Walters | Northeastern University;The Robotics and AI Institute;Northeastern University;Northeastern University;Northeastern University;Northeastern University;Brown University;Northeastern University;Northeastern University | Poster | main | Manipulation policy learning;Generative model;Geometric learning | https://openreview.net/forum?id=56IzghzjfZ | 7 | IMAGINATION POLICY: Using Generative Point Cloud Models for Learning Manipulation Policies
Humans can imagine goal states during planning and perform actions to match those goals. In this work, we propose IMAGINATION POLICY, a novel multi-task key-frame policy network for solving high-precision pick and place tasks. In... | [
-0.011953826993703842,
-0.01901201345026493,
0.0062805903144180775,
0.01804577000439167,
-0.032465819269418716,
-0.008374120108783245,
0.013748281635344028,
0.025195980444550514,
0.0065704635344445705,
0.02639228291809559,
-0.017134739086031914,
-0.016831062734127045,
-0.018100982531905174,
... | ||
corl_2024_5Awumz1VKU | 5Awumz1VKU | corl | 2,024 | Learning Differentiable Tensegrity Dynamics using Graph Neural Networks | Tensegrity robots are composed of rigid struts and flexible cables. They constitute an emerging class of hybrid rigid-soft robotic systems and are promising systems for a wide array of applications, ranging from locomotion to assembly. They are difficult to control and model accurately, however, due to their compliance... | Nelson Chen;Kun Wang;William R. Johson III;Rebecca Kramer-Bottiglio;Kostas Bekris;Mridul Aanjaneya | Rutgers University;;;Yale University;Rutgers University;Rutgers University, New Brunswick | Poster | main | graph neural networks;differentiable simulation;tensegrity robots | https://github.com/nchen9191/tensegrity_gnn_simulator_public | https://openreview.net/forum?id=5Awumz1VKU | 0 | Learning Differentiable Tensegrity Dynamics using Graph Neural Networks
Tensegrity robots are composed of rigid struts and flexible cables. They constitute an emerging class of hybrid rigid-soft robotic systems and are promising systems for a wide array of applications, ranging from locomotion to assembly. They are dif... | [
-0.03626123070716858,
-0.021227890625596046,
-0.006162190344184637,
-0.008034423924982548,
-0.0035757345613092184,
-0.07000765949487686,
0.00605124281719327,
-0.008986719883978367,
0.02315097488462925,
0.02500009350478649,
-0.04323241114616394,
-0.010845084674656391,
-0.0029077401850372553,
... | |
corl_2024_5W0iZR9J7h | 5W0iZR9J7h | corl | 2,024 | DexGraspNet 2.0: Learning Generative Dexterous Grasping in Large-scale Synthetic Cluttered Scenes | Grasping in cluttered scenes remains highly challenging for dexterous hands due to the scarcity of data. To address this problem, we present a large-scale synthetic dataset, encompassing 1319 objects, 8270 scenes, and 426 million grasps. Beyond benchmarking, we also explore data-efficient learning strategies from grasp... | Jialiang Zhang;Haoran Liu;Danshi Li;XinQiang Yu;Haoran Geng;Yufei Ding;Jiayi Chen;He Wang | Peking University;Peking University;New York University;Institute of Computing Technology, Chinese Academy of Sciences;Peking University;Peking University;Peking University;Peking University | Poster | main | Dexterous Grasping;Synthetic Data;Generative Models | https://openreview.net/forum?id=5W0iZR9J7h | 10 | DexGraspNet 2.0: Learning Generative Dexterous Grasping in Large-scale Synthetic Cluttered Scenes
Grasping in cluttered scenes remains highly challenging for dexterous hands due to the scarcity of data. To address this problem, we present a large-scale synthetic dataset, encompassing 1319 objects, 8270 scenes, and 426 ... | [
-0.045684266835451126,
-0.009716851636767387,
-0.033613529056310654,
-0.017691826447844505,
-0.027606403455138206,
-0.058263447135686874,
-0.022032396867871284,
0.025290178135037422,
-0.014048006385564804,
0.004453557077795267,
-0.03088301606476307,
0.001333477790467441,
0.000898008467629551... | ||
corl_2024_5iXG6EgByK | 5iXG6EgByK | corl | 2,024 | Promptable Closed-loop Traffic Simulation | Simulation stands as a cornerstone for safe and efficient autonomous driving development. At its core a simulation system ought to produce realistic, reactive, and controllable traffic patterns. In this paper, we propose ProSim, a multimodal promptable closed-loop traffic simulation framework. ProSim allows the user to... | Shuhan Tan;Boris Ivanovic;Yuxiao Chen;Boyi Li;Xinshuo Weng;Yulong Cao;Philipp Kraehenbuehl;Marco Pavone | NVIDIA;NVIDIA;University of California, Berkeley;NVIDIA;NVIDIA;Apple;Stanford University;California Institute of Technology | Poster | main | Autonomous Driving;Scenario Generation;Traffic Simulation | https://openreview.net/forum?id=5iXG6EgByK | 5 | Promptable Closed-loop Traffic Simulation
Simulation stands as a cornerstone for safe and efficient autonomous driving development. At its core a simulation system ought to produce realistic, reactive, and controllable traffic patterns. In this paper, we propose ProSim, a multimodal promptable closed-loop traffic simul... | [
-0.06938907504081726,
-0.04614853858947754,
0.013532347977161407,
-0.028265517204999924,
-0.030242256820201874,
-0.0001450512936571613,
0.0027780570089817047,
-0.00005455660721054301,
0.004826383199542761,
0.041419144719839096,
-0.03353066369891167,
-0.010539527982473373,
0.00786076951771974... | ||
corl_2024_5lSkn5v4LK | 5lSkn5v4LK | corl | 2,024 | EquiGraspFlow: SE(3)-Equivariant 6-DoF Grasp Pose Generative Flows | Traditional methods for synthesizing 6-DoF grasp poses from 3D observations often rely on geometric heuristics, resulting in poor generalizability, limited grasp options, and higher failure rates. Recently, data-driven methods have been proposed that use generative models to learn the distribution of grasp poses and ge... | Byeongdo Lim;Jongmin Kim;Jihwan Kim;Yonghyeon Lee;Frank C. Park | Seoul National University;Seoul National University;Seoul National University;Korea Institute for Advanced Study;Seoul National University | Poster | main | 6-DoF grasp pose generation;equivariance;generative models;continuous normalizing flows | https://github.com/bdlim99/EquiGraspFlow | https://openreview.net/forum?id=5lSkn5v4LK | 7 | EquiGraspFlow: SE(3)-Equivariant 6-DoF Grasp Pose Generative Flows
Traditional methods for synthesizing 6-DoF grasp poses from 3D observations often rely on geometric heuristics, resulting in poor generalizability, limited grasp options, and higher failure rates. Recently, data-driven methods have been proposed that us... | [
-0.06999378651380539,
-0.02757769636809826,
-0.03995508700609207,
-0.03074442595243454,
-0.012594538740813732,
-0.035304516553878784,
-0.0001327482023043558,
0.05714590847492218,
-0.01826750859618187,
0.0082154031842947,
-0.01804131455719471,
0.003558047581464052,
-0.006953234784305096,
0.... | |
corl_2024_5u9l6U61S7 | 5u9l6U61S7 | corl | 2,024 | GenSim2: Scaling Robot Data Generation with Multi-modal and Reasoning LLMs | Robotic simulation today remains challenging to scale up due to the human efforts required to create diverse simulation tasks and scenes. Simulation-trained policies also face scalability issues as many sim-to-real methods focus on a single task. To address these challenges, this work proposes GenSim2, a scalable frame... | Pu Hua;Minghuan Liu;Annabella Macaluso;Yunfeng Lin;Weinan Zhang;Huazhe Xu;Lirui Wang | Electronic Engineering, Tsinghua University;Shanghai Jiaotong University;University of California, San Diego;;Shanghai Jiaotong University;Tsinghua University;Massachusetts Institute of Technology | Poster | main | Generative Simulation; Robotics; Learning | https://github.com/GenSim2/GenSim2 | https://openreview.net/forum?id=5u9l6U61S7 | 11 | GenSim2: Scaling Robot Data Generation with Multi-modal and Reasoning LLMs
Robotic simulation today remains challenging to scale up due to the human efforts required to create diverse simulation tasks and scenes. Simulation-trained policies also face scalability issues as many sim-to-real methods focus on a single task... | [
-0.069534070789814,
-0.012155462056398392,
0.0076470766216516495,
0.014059105888009071,
-0.047136079519987106,
-0.012573335319757462,
-0.04464741051197052,
0.023215169087052345,
0.007154915016144514,
0.017114222049713135,
-0.04624461382627487,
0.012025456875562668,
-0.002484023105353117,
0... | |
corl_2024_67tTQeO4HQ | 67tTQeO4HQ | corl | 2,024 | In-Flight Attitude Control of a Quadruped using Deep Reinforcement Learning | We present the development and real world demonstration of an in-flight attitude control law for a small low-cost quadruped with a five-bar-linkage leg design using only its legs as reaction masses. The control law is trained using deep reinforcement learning (DRL) and specifically through Proximal Policy Optimization ... | Tarek El-Agroudi;Finn Gross Maurer;Jørgen Anker Olsen;Kostas Alexis | Norwegian University of Science and Technology;Norwegian University of Science and Technology;Norwegian University of Science and Technology;Norwegian University of Science and Technology | Poster | main | Deep Reinforcement Learning;Legged Robotics | https://github.com/ntnu-arl/Eurepus-RL and https://github.com/ntnu-arl/Eurepus-design | https://openreview.net/forum?id=67tTQeO4HQ | 1 | In-Flight Attitude Control of a Quadruped using Deep Reinforcement Learning
We present the development and real world demonstration of an in-flight attitude control law for a small low-cost quadruped with a five-bar-linkage leg design using only its legs as reaction masses. The control law is trained using deep reinfor... | [
-0.05853114277124405,
-0.021699439734220505,
-0.004046677611768246,
-0.014392297714948654,
-0.04184495657682419,
0.0183048564940691,
0.018526844680309296,
0.030745498836040497,
0.004358849488198757,
0.03198494017124176,
-0.03962506353855133,
-0.012819874100387096,
0.0187210850417614,
0.018... | |
corl_2024_6FGlpzC9Po | 6FGlpzC9Po | corl | 2,024 | Steering Your Generalists: Improving Robotic Foundation Models via Value Guidance | Large, general-purpose robotic policies trained on diverse demonstration datasets have been shown to be remarkably effective both for controlling a variety of robots in a range of different scenes, and for acquiring broad repertoires of manipulation skills. However, the data that such policies are trained on is general... | Mitsuhiko Nakamoto;Oier Mees;Aviral Kumar;Sergey Levine | University of California, Berkeley;Electrical Engineering & Computer Science Department, University of California, Berkeley;Google DeepMind;Google | Poster | main | generalist policies;value functions;robot reinforcement learning | https://openreview.net/forum?id=6FGlpzC9Po | 7 | Steering Your Generalists: Improving Robotic Foundation Models via Value Guidance
Large, general-purpose robotic policies trained on diverse demonstration datasets have been shown to be remarkably effective both for controlling a variety of robots in a range of different scenes, and for acquiring broad repertoires of m... | [
-0.045459598302841187,
-0.0627002939581871,
-0.0339532271027565,
0.012251456268131733,
-0.033915501087903976,
-0.026332613080739975,
0.005192014388740063,
-0.005833352915942669,
0.021069863811135292,
0.02355976589024067,
-0.042252905666828156,
0.006276631262153387,
0.015533600933849812,
0.... | ||
corl_2024_6X3ybeVpDi | 6X3ybeVpDi | corl | 2,024 | Online Transfer and Adaptation of Tactile Skill: A Teleoperation Framework | This paper presents a teleoperation framework designed for online learning and adaptation of tactile skills, which provides an intuitive interface without need for physical access to execution robot. The proposed tele-teaching approach utilizes periodical Dynamical Movement Primitives (DMP) and Recursive Least Square (... | Xiao Chen;Tianle Ni;Kübra Karacan;Hamid Sadeghian;Sami Haddadin | Technische Universität München;Technische Universität München;Technische Universität München;; | Poster | main | Learning from Demonstration;Online Adaptation;Tactile Skill;Teleoperation;Autonomy Allocation | https://openreview.net/forum?id=6X3ybeVpDi | 1 | Online Transfer and Adaptation of Tactile Skill: A Teleoperation Framework
This paper presents a teleoperation framework designed for online learning and adaptation of tactile skills, which provides an intuitive interface without need for physical access to execution robot. The proposed tele-teaching approach utilizes ... | [
-0.07193850725889206,
0.006770683452486992,
-0.0006140531622804701,
0.032565515488386154,
-0.0016294256784021854,
0.013458573259413242,
0.05030176043510437,
-0.010855171829462051,
0.03252871707081795,
0.023586999624967575,
-0.02794746682047844,
0.012768625281751156,
0.013237789273262024,
0... | ||
corl_2024_6oESa4g05O | 6oESa4g05O | corl | 2,024 | Distribution Discrepancy and Feature Heterogeneity for Active 3D Object Detection | LiDAR-based 3D object detection is a critical technology for the development of autonomous driving and robotics. However, the high cost of data annotation limits its advancement. We propose a novel and effective active learning (AL) method called Distribution Discrepancy and Feature Heterogeneity (DDFH), which simultan... | Huang-Yu Chen;Jia-Fong Yeh;Jiawei;Pin-Hsuan Peng;Winston H. Hsu | National Taiwan University;Sony Group Corporation;National Taiwan University;National Taiwan University;National Taiwan University | Poster | main | Active Learning;LiDAR 3D Object Detection;Autonomous Driving | https://github.com/Coolshanlan/DDFH-active-3Ddet | https://openreview.net/forum?id=6oESa4g05O | 0 | Distribution Discrepancy and Feature Heterogeneity for Active 3D Object Detection
LiDAR-based 3D object detection is a critical technology for the development of autonomous driving and robotics. However, the high cost of data annotation limits its advancement. We propose a novel and effective active learning (AL) metho... | [
-0.0291119571775198,
-0.07518510520458221,
-0.021860692650079727,
0.013068311847746372,
-0.01922386884689331,
-0.0023606696631759405,
0.05726183205842972,
0.008413783274590969,
0.023838309571146965,
0.01931295171380043,
-0.05665607377886772,
-0.05131116136908531,
0.0058393171057105064,
0.0... | |
corl_2024_7E3JAys1xO | 7E3JAys1xO | corl | 2,024 | D$^3$RoMa: Disparity Diffusion-based Depth Sensing for Material-Agnostic Robotic Manipulation | Depth sensing is an important problem for 3D vision-based robotics. Yet, a real-world active stereo or ToF depth camera often produces noisy and incomplete depth which bottlenecks robot performances. In this work, we propose D3RoMa, a learning-based depth estimation framework on stereo image pairs that predicts clean a... | Songlin Wei;Haoran Geng;Jiayi Chen;Congyue Deng;Cui Wenbo;Chengyang Zhao;Xiaomeng Fang;Leonidas Guibas;He Wang | Peking University;Peking University;Stanford University;University of Chinese Academy of Sciences;Peking University;;Stanford University;Peking University;Peking University | Poster | main | Depth Estimation;Diffusion Model;Stereo Vision | https://openreview.net/forum?id=7E3JAys1xO | 4 | D$^3$RoMa: Disparity Diffusion-based Depth Sensing for Material-Agnostic Robotic Manipulation
Depth sensing is an important problem for 3D vision-based robotics. Yet, a real-world active stereo or ToF depth camera often produces noisy and incomplete depth which bottlenecks robot performances. In this work, we propose D... | [
-0.06711392849683762,
-0.006707737687975168,
-0.03746829181909561,
-0.0024857011158019304,
-0.015581325627863407,
0.008837033063173294,
-0.00551514932885766,
0.0184599868953228,
0.029024217277765274,
0.03712102398276329,
-0.0352201908826828,
0.012547307647764683,
0.007027589250355959,
-0.0... | ||
corl_2024_7c5rAY8oU3 | 7c5rAY8oU3 | corl | 2,024 | Automated Creation of Digital Cousins for Robust Policy Learning | Training robot policies in the real world can be unsafe, costly, and difficult to scale. Simulation serves as an inexpensive and potentially limitless source of training data, but suffers from the semantics and physics disparity between simulated and real-world environments. These discrepancies can be minimized by trai... | Tianyuan Dai;Josiah Wong;Yunfan Jiang;Chen Wang;Cem Gokmen;Ruohan Zhang;Jiajun Wu;Li Fei-Fei | ;Stanford University;Stanford University;Computer Science Department, Stanford University;Stanford University;Stanford University;Stanford University;Stanford University | Poster | main | Real-to-Sim; Digital Twin; Sim-to-Real Transfer | https://github.com/cremebrule/digital-cousins | https://openreview.net/forum?id=7c5rAY8oU3 | 11 | Automated Creation of Digital Cousins for Robust Policy Learning
Training robot policies in the real world can be unsafe, costly, and difficult to scale. Simulation serves as an inexpensive and potentially limitless source of training data, but suffers from the semantics and physics disparity between simulated and real... | [
-0.07176607847213745,
-0.018415160477161407,
-0.022469528019428253,
0.024761950597167015,
-0.026542840525507927,
0.006422571372240782,
-0.005394770763814449,
0.049561791121959686,
0.028854208067059517,
0.059148285537958145,
-0.015933282673358917,
0.014057664200663567,
0.0009384011500515044,
... | |
corl_2024_7ddT4eklmQ | 7ddT4eklmQ | corl | 2,024 | ACE: A Cross-platform and visual-Exoskeletons System for Low-Cost Dexterous Teleoperation | Bimanual robotic manipulation with dexterous hands has a large potential workability and a wide workspace as it follows the most natural human workflow.
Learning from human demonstrations has proven highly effective for learning a dexterous manipulation policy. To collect such data, teleoperation serves as a straightfo... | Shiqi Yang;Minghuan Liu;Yuzhe Qin;Runyu Ding;Jialong Li;Xuxin Cheng;Ruihan Yang;Sha Yi;Xiaolong Wang | University of California, San Diego;Shanghai Jiaotong University;University of California, San Diego;Electrical and Electronic Engineering, University of Hong Kong;University of California, San Diego;University of California, San Diego;University of California, San Diego;Carnegie Mellon University;University of Califor... | Poster | main | Teleopration System; Hardware; Imitation Learning; Robot Learning; Exoskeletons | https://github.com/ACETeleop/ACETeleop | https://openreview.net/forum?id=7ddT4eklmQ | 37 | ACE: A Cross-platform and visual-Exoskeletons System for Low-Cost Dexterous Teleoperation
Bimanual robotic manipulation with dexterous hands has a large potential workability and a wide workspace as it follows the most natural human workflow.
Learning from human demonstrations has proven highly effective for learning a... | [
-0.02170124277472496,
-0.04018063098192215,
-0.0462910458445549,
0.025571174919605255,
0.0192570760846138,
0.010989494621753693,
0.004795752465724945,
0.018525676801800728,
-0.0032033405732363462,
0.04580961912870407,
-0.0028561577200889587,
-0.002186094643548131,
0.0290707778185606,
0.005... | |
corl_2024_7vzDBvviRO | 7vzDBvviRO | corl | 2,024 | UBSoft: A Simulation Platform for Robotic Skill Learning in Unbounded Soft Environments | It is desired to equip robots with the capability of interacting with various soft materials as they are ubiquitous in the real world. While physics simulations are one of the predominant methods for data collection and robot training, simulating soft materials presents considerable challenges. Specifically, it is sign... | Chunru Lin;Jugang Fan;Yian Wang;Zeyuan Yang;Zhehuan Chen;Lixing Fang;Tsun-Hsuan Wang;Zhou Xian;Chuang Gan | University of Massachusetts at Amherst;South China University of Technology;, Tsinghua University;University of Massachusetts at Amherst;Tsinghua University;Liquid AI;Carnegie Mellon University;University of Massachusetts at Amherst;NVIDIA | Poster | main | Soft-Body Manipulation;Locomotion;Physics Simulation | https://openreview.net/forum?id=7vzDBvviRO | 1 | UBSoft: A Simulation Platform for Robotic Skill Learning in Unbounded Soft Environments
It is desired to equip robots with the capability of interacting with various soft materials as they are ubiquitous in the real world. While physics simulations are one of the predominant methods for data collection and robot traini... | [
-0.05729326978325844,
-0.009101850911974907,
-0.04483179748058319,
-0.03314456343650818,
-0.01432331744581461,
0.0058159674517810345,
-0.015245024114847183,
0.010120335966348648,
0.043652016669511795,
0.024941375479102135,
-0.007889806292951107,
0.024554258212447166,
0.03386349231004715,
0... | ||
corl_2024_7wMlwhCvjS | 7wMlwhCvjS | corl | 2,024 | GenDP: 3D Semantic Fields for Category-Level Generalizable Diffusion Policy | Diffusion-based policies have shown remarkable capability in executing complex robotic manipulation tasks but lack explicit characterization of geometry and semantics, which often limits their ability to generalize to unseen objects and layouts. To enhance the generalization capabilities of Diffusion Policy, we introdu... | Yixuan Wang;Guang Yin;Binghao Huang;Tarik Kelestemur;Jiuguang Wang;Yunzhu Li | University of Illinois, Urbana Champaign;University of Illinois, Urbana Champaign;University of Illinois Urbana-Champaign;Boston Dynamics AI Institute;;University of Illinois Urbana-Champaign | Poster | main | Semantic Fields;Category-Level Generalization;Imitation Learning;Diffusion Models | https://github.com/WangYixuan12/gendp | https://openreview.net/forum?id=7wMlwhCvjS | 14 | GenDP: 3D Semantic Fields for Category-Level Generalizable Diffusion Policy
Diffusion-based policies have shown remarkable capability in executing complex robotic manipulation tasks but lack explicit characterization of geometry and semantics, which often limits their ability to generalize to unseen objects and layouts... | [
-0.07962880283594131,
-0.012606070376932621,
0.0007941459771245718,
-0.009352008812129498,
-0.024683469906449318,
-0.015304116532206535,
-0.017336765304207802,
0.0007799037848599255,
0.023042766377329826,
0.0060478150844573975,
-0.027983106672763824,
0.026925764977931976,
0.00261601037345826... | |
corl_2024_7yMZAUkXa4 | 7yMZAUkXa4 | corl | 2,024 | MimicTouch: Leveraging Multi-modal Human Tactile Demonstrations for Contact-rich Manipulation | Tactile sensing is critical to fine-grained, contact-rich manipulation tasks, such as insertion and assembly. Prior research has shown the possibility of learning tactile-guided policy from teleoperated demonstration data. However, to provide the demonstration, human users often rely on visual feedback to control the r... | Kelin Yu;Yunhai Han;Qixian Wang;Vaibhav Saxena;Danfei Xu;Ye Zhao | Georgia Institute of Technology;Georgia Institute of Technology;;Georgia Institute of Technology;NVIDIA;Georgia Institute of Technology | Poster | main | Tactile Sensing;Learning from Human;Data Collection;Imitation Learning;Reinforcement Learning | https://openreview.net/forum?id=7yMZAUkXa4 | 16 | MimicTouch: Leveraging Multi-modal Human Tactile Demonstrations for Contact-rich Manipulation
Tactile sensing is critical to fine-grained, contact-rich manipulation tasks, such as insertion and assembly. Prior research has shown the possibility of learning tactile-guided policy from teleoperated demonstration data. How... | [
-0.041096411645412445,
-0.03896057605743408,
-0.010946154594421387,
0.016828907653689384,
-0.02021678350865841,
-0.03437589481472969,
0.008999045938253403,
0.002058733254671097,
-0.011664236895740032,
0.044263340532779694,
-0.0010356958955526352,
-0.013431824743747711,
0.018891092389822006,
... | ||
corl_2024_82bpTugrMt | 82bpTugrMt | corl | 2,024 | Monocular Event-Based Vision for Obstacle Avoidance with a Quadrotor | We present the first static-obstacle avoidance method for quadrotors using just an onboard, monocular event camera. Quadrotors are capable of fast and agile flight in cluttered environments when piloted manually, but vision-based autonomous flight in unknown environments is difficult in part due to the sensor limitatio... | Anish Bhattacharya;Marco Cannici;Nishanth Rao;Yuezhan Tao;Vijay Kumar;Nikolai Matni;Davide Scaramuzza | University of Pennsylvania;University of Zurich;;University of Pennsylvania;University of Pennsylvania;School of Engineering and Applied Science, University of Pennsylvania; | Poster | main | event-based vision;learning for control;simulation-to-real transfer;aerial robotics | https://openreview.net/forum?id=82bpTugrMt | 5 | Monocular Event-Based Vision for Obstacle Avoidance with a Quadrotor
We present the first static-obstacle avoidance method for quadrotors using just an onboard, monocular event camera. Quadrotors are capable of fast and agile flight in cluttered environments when piloted manually, but vision-based autonomous flight in ... | [
-0.003234506817534566,
-0.03464537113904953,
-0.016393175348639488,
-0.025631941854953766,
0.0035584268625825644,
0.01089122798293829,
-0.005356418434530497,
0.038194410502910614,
0.0021899817511439323,
0.03554671257734299,
-0.028166968375444412,
-0.017942357808351517,
0.010177664458751678,
... | ||
corl_2024_8Ar8b00GJC | 8Ar8b00GJC | corl | 2,024 | Autonomous Improvement of Instruction Following Skills via Foundation Models | Intelligent robots capable of improving from autonomously collected experience have the potential to transform robot learning: instead of collecting costly teleoperated demonstration data, large-scale deployment of fleets of robots can quickly collect larger quantities of autonomous data useful for training better robo... | Zhiyuan Zhou;Pranav Atreya;Abraham Lee;Homer Rich Walke;Oier Mees;Sergey Levine | University of California, Berkeley;University of California, Berkeley;University of California, Berkeley;University of California, Berkeley;Electrical Engineering & Computer Science Department, University of California, Berkeley;Google | Poster | main | Autonomous Improvement;Instruction Following Skills;Scaled Data Collection | https://github.com/rail-berkeley/soar | https://openreview.net/forum?id=8Ar8b00GJC | 12 | Autonomous Improvement of Instruction Following Skills via Foundation Models
Intelligent robots capable of improving from autonomously collected experience have the potential to transform robot learning: instead of collecting costly teleoperated demonstration data, large-scale deployment of fleets of robots can quickly... | [
-0.029551200568675995,
-0.03472913056612015,
0.006278242915868759,
0.011557884514331818,
-0.02209867537021637,
-0.032454539090394974,
0.006698950193822384,
0.0031067593954503536,
0.009819579310715199,
0.01484032440930605,
-0.007692927960306406,
-0.03596813604235649,
0.004114606883376837,
0... | |
corl_2024_8JLmTZsxGh | 8JLmTZsxGh | corl | 2,024 | Learning Performance-oriented Control Barrier Functions Under Complex Safety Constraints and Limited Actuation | Control Barrier Functions (CBFs) offer an elegant framework for constraining nonlinear control system dynamics to an invariant subset of a pre-specified safe set. However, finding a CBF that simultaneously promotes performance by maximizing the resulting control invariant set while accommodating complex safety constrai... | Lakshmideepakreddy Manda;Shaoru Chen;Mahyar Fazlyab | Whiting School of Engineering;Microsoft Research;Johns Hopkins University | Poster | main | Control Barrier Functions;Safety;Hamilton-Jacobi Partial Differential Equation | https://github.com/o4lc/PINN-CBF | https://openreview.net/forum?id=8JLmTZsxGh | 8 | Learning Performance-oriented Control Barrier Functions Under Complex Safety Constraints and Limited Actuation
Control Barrier Functions (CBFs) offer an elegant framework for constraining nonlinear control system dynamics to an invariant subset of a pre-specified safe set. However, finding a CBF that simultaneously pro... | [
-0.01760326512157917,
-0.01763986237347126,
-0.08710139989852905,
-0.0031725221779197454,
0.03304729610681534,
-0.00009006348409457132,
0.025233788415789604,
0.029332676902413368,
0.012662640772759914,
0.026880662888288498,
-0.04464861750602722,
-0.014153976924717426,
0.020878717303276062,
... | |
corl_2024_8LPXeGhhbH | 8LPXeGhhbH | corl | 2,024 | RAM: Retrieval-Based Affordance Transfer for Generalizable Zero-Shot Robotic Manipulation | This work proposes a retrieve-and-transfer framework for zero-shot robotic manipulation, dubbed RAM, featuring generalizability across various objects, environments, and embodiments. Unlike existing approaches that learn manipulation from expensive in-domain demonstrations, RAM capitalizes on a retrieval-based affordan... | Yuxuan Kuang;Junjie Ye;Haoran Geng;Jiageng Mao;Congyue Deng;Leonidas Guibas;He Wang;Yue Wang | Peking University;University of Southern California;Peking University;;Stanford University;Stanford University;Peking University;NVIDIA | Poster | main | Hierarchical Retrieval;Affordance Transfer;Zero-Shot Robotic Manipulation;Visual Foundation Models | https://github.com/yxKryptonite/RAM_code | https://openreview.net/forum?id=8LPXeGhhbH | 26 | RAM: Retrieval-Based Affordance Transfer for Generalizable Zero-Shot Robotic Manipulation
This work proposes a retrieve-and-transfer framework for zero-shot robotic manipulation, dubbed RAM, featuring generalizability across various objects, environments, and embodiments. Unlike existing approaches that learn manipulat... | [
-0.007961239665746689,
0.00681333988904953,
-0.009581259451806545,
0.025864772498607635,
-0.015552189201116562,
-0.00617458950728178,
0.008947137743234634,
0.035140540450811386,
0.012571352533996105,
0.029882419854402542,
-0.04487917572259903,
0.011775229126214981,
0.012238091789186,
0.045... | |
corl_2024_8PcRynpd1m | 8PcRynpd1m | corl | 2,024 | Safe Bayesian Optimization for the Control of High-Dimensional Embodied Systems | Learning to move is a primary goal for animals and robots, where ensuring safety is often important when optimizing control policies on the embodied systems. For complex tasks such as the control of human or humanoid control, the high-dimensional parameter space adds complexity to the safe optimization effort. Current ... | Yunyue Wei;Zeji Yi;Hongda Li;Saraswati Soedarmadji;Yanan Sui | Tsinghua University;Tsinghua University;Tsinghua University;California Institute of Technology;Tsinghua University | Poster | main | Safe Bayesian Optimization;High-dimensional Embodied System | https://openreview.net/forum?id=8PcRynpd1m | 0 | Safe Bayesian Optimization for the Control of High-Dimensional Embodied Systems
Learning to move is a primary goal for animals and robots, where ensuring safety is often important when optimizing control policies on the embodied systems. For complex tasks such as the control of human or humanoid control, the high-dimen... | [
-0.020048297941684723,
-0.027521364390850067,
-0.015562563203275204,
0.028223149478435516,
-0.01579965278506279,
-0.014054672792553902,
0.02321581542491913,
-0.014064156450331211,
0.012025185860693455,
0.05344948545098305,
-0.06259166449308395,
0.005159070249646902,
0.05618075653910637,
0.... | ||
corl_2024_8XFT1PatHy | 8XFT1PatHy | corl | 2,024 | Splat-MOVER: Multi-Stage, Open-Vocabulary Robotic Manipulation via Editable Gaussian Splatting | We present Splat-MOVER, a modular robotics stack for open-vocabulary
robotic manipulation, which leverages the editability of Gaussian Splatting (GSplat)
scene representations to enable multi-stage manipulation tasks. Splat-MOVER
consists of: (i) ASK-Splat, a GSplat representation that distills semantic and grasp
affor... | Olaolu Shorinwa;Johnathan Tucker;Aliyah Smith;Aiden Swann;Timothy Chen;Roya Firoozi;Monroe David Kennedy;Mac Schwager | Stanford University;;;Stanford University;Stanford University;;Stanford University;Stanford University | Poster | main | Gaussian Splatting;Robotic Grasping;Robotic Manipulation;Scene Editing | https://splatmover.github.io | https://openreview.net/forum?id=8XFT1PatHy | 23 | Splat-MOVER: Multi-Stage, Open-Vocabulary Robotic Manipulation via Editable Gaussian Splatting
We present Splat-MOVER, a modular robotics stack for open-vocabulary
robotic manipulation, which leverages the editability of Gaussian Splatting (GSplat)
scene representations to enable multi-stage manipulation tasks. Splat-M... | [
-0.04607889801263809,
-0.023553885519504547,
-0.005222458858042955,
0.01498299092054367,
-0.00491012167185545,
0.021147053688764572,
-0.01934652216732502,
0.0019934452138841152,
0.014523671939969063,
0.04982694238424301,
-0.016259897500276566,
0.039868902415037155,
0.022690366953611374,
0.... | |
corl_2024_8Yu0TNJNGK | 8Yu0TNJNGK | corl | 2,024 | AnyRotate: Gravity-Invariant In-Hand Object Rotation with Sim-to-Real Touch | Human hands are capable of in-hand manipulation in the presence of different hand motions. For a robot hand, harnessing rich tactile information to achieve this level of dexterity still remains a significant challenge. In this paper, we present AnyRotate, a system for gravity-invariant multi-axis in-hand object rotatio... | Max Yang;chenghua lu;Alex Church;Yijiong Lin;Christopher J. Ford;Haoran Li;Efi Psomopoulou;David A.W. Barton;Nathan F. Lepora | University of Bristol;University of Bristol;;University of Bristol;University of Bristol;University of Bristol;University of Bristol;University of Bristol;University of Bristol | Poster | main | Tactile Sensing;In-hand Object Rotation;Reinforcement Learning | https://openreview.net/forum?id=8Yu0TNJNGK | 19 | AnyRotate: Gravity-Invariant In-Hand Object Rotation with Sim-to-Real Touch
Human hands are capable of in-hand manipulation in the presence of different hand motions. For a robot hand, harnessing rich tactile information to achieve this level of dexterity still remains a significant challenge. In this paper, we present... | [
-0.06469301134347916,
-0.003942988812923431,
-0.005412860307842493,
0.031039955094456673,
-0.03313043713569641,
-0.046998560428619385,
0.026858987286686897,
0.024600517004728317,
0.022136732935905457,
0.0015596970915794373,
-0.02908012643456459,
0.03830065205693245,
-0.0013998777139931917,
... | ||
corl_2024_928V4Umlys | 928V4Umlys | corl | 2,024 | DriveVLM: The Convergence of Autonomous Driving and Large Vision-Language Models | A primary hurdle of autonomous driving in urban environments is understanding complex and long-tail scenarios, such as challenging road conditions and delicate human behaviors. We introduce DriveVLM, an autonomous driving system leveraging Vision-Language Models (VLMs) for enhanced scene understanding and planning capa... | Xiaoyu Tian;Junru Gu;Bailin Li;Yicheng Liu;Yang Wang;Zhiyong Zhao;Kun Zhan;Peng Jia;XianPeng Lang;Hang Zhao | IIIS, Tsinghua University;Tsinghua University;Li Auto;Tsinghua University;LiAuto;LiAuto;Li Auto;LiAuto;Tsinghua University;Li Auto | Poster | main | Autonomous Driving;Vision Language Model;Dual System | https://openreview.net/forum?id=928V4Umlys | 190 | DriveVLM: The Convergence of Autonomous Driving and Large Vision-Language Models
A primary hurdle of autonomous driving in urban environments is understanding complex and long-tail scenarios, such as challenging road conditions and delicate human behaviors. We introduce DriveVLM, an autonomous driving system leveraging... | [
-0.06513936072587967,
-0.01409372128546238,
0.026219148188829422,
-0.024306561797857285,
-0.05422090366482735,
0.048353161662817,
0.01169834565371275,
0.021669792011380196,
0.024380836635828018,
0.022876763716340065,
-0.014316546730697155,
-0.03386949375271797,
0.030601385980844498,
0.0106... | ||
corl_2024_97QXO0uBEO | 97QXO0uBEO | corl | 2,024 | Handling Long-Term Safety and Uncertainty in Safe Reinforcement Learning | Safety is one of the key issues preventing the deployment of reinforcement learning techniques in real-world robots. While most approaches in the Safe Reinforcement Learning area do not require prior knowledge of constraints and robot kinematics and rely solely on data, it is often difficult to deploy them in complex r... | Jonas Günster;Puze Liu;Jan Peters;Davide Tateo | Technische Universität Darmstadt;TU Darmstadt;TU Darmstadt;Technische Universität Darmstadt | Poster | main | Safe Reinforcement Learning;Chance Constraint;Distributional RL | https://github.com/cube1324/d-atacom | https://openreview.net/forum?id=97QXO0uBEO | 2 | Handling Long-Term Safety and Uncertainty in Safe Reinforcement Learning
Safety is one of the key issues preventing the deployment of reinforcement learning techniques in real-world robots. While most approaches in the Safe Reinforcement Learning area do not require prior knowledge of constraints and robot kinematics a... | [
-0.042707160115242004,
-0.015853989869356155,
-0.015133354812860489,
-0.00014289747923612595,
-0.048699818551540375,
0.010515594854950905,
0.0022626544814556837,
0.021884575486183167,
-0.024956759065389633,
0.017314225435256958,
-0.013910169713199139,
0.0016012814594432712,
-0.00760934641584... | |
corl_2024_9HkElMlPbU | 9HkElMlPbU | corl | 2,024 | Contrastive Imitation Learning for Language-guided Multi-Task Robotic Manipulation | Developing robots capable of executing various manipulation tasks, guided by natural language instructions and visual observations of intricate real-world environments, remains a significant challenge in robotics. Such robot agents need to understand linguistic commands and distinguish between the requirements of diff... | Teli Ma;Jiaming Zhou;Zifan Wang;Ronghe Qiu;Junwei Liang | Hong Kong University of Science and Technology;Hong Kong University of Science and Technology (Guangzhou);the Hong Kong University of Science and Technology（Guangzhou);Hong Kong University of Science and Technology;Hong Kong University of Science and Technology | Poster | main | Contrastive imitation learning;Multi-task learning;Robotic manipulation | https://openreview.net/forum?id=9HkElMlPbU | 11 | Contrastive Imitation Learning for Language-guided Multi-Task Robotic Manipulation
Developing robots capable of executing various manipulation tasks, guided by natural language instructions and visual observations of intricate real-world environments, remains a significant challenge in robotics. Such robot agents need ... | [
-0.02258148230612278,
-0.01576460525393486,
0.0237437654286623,
0.0025344209279865026,
0.005428595934063196,
-0.01434403844177723,
0.002077810000628233,
0.0013306286418810487,
-0.00007184990681707859,
0.004206354729831219,
-0.03955449163913727,
-0.04475708678364754,
-0.02876187115907669,
0... | ||
corl_2024_9XV3dBqcfe | 9XV3dBqcfe | corl | 2,024 | Generalized Animal Imitator: Agile Locomotion with Versatile Motion Prior | The agility of animals, particularly in complex activities such as running, turning, jumping, and backflipping, stands as an exemplar for robotic system design. Transferring this suite of behaviors to legged robotic systems introduces essential inquiries: How can a robot be trained to learn multiple locomotion behavior... | Ruihan Yang;Zhuoqun Chen;Jianhan Ma;Chongyi Zheng;Yiyu Chen;Quan Nguyen;Xiaolong Wang | University of California, San Diego;University of California, San Diego;University of California, San Diego;Princeton University;University of Southern California;University of Southern California;University of California, San Diego | Poster | main | Legged Robots;Imitation Learning;Agile Locomotion | https://openreview.net/forum?id=9XV3dBqcfe | 19 | Generalized Animal Imitator: Agile Locomotion with Versatile Motion Prior
The agility of animals, particularly in complex activities such as running, turning, jumping, and backflipping, stands as an exemplar for robotic system design. Transferring this suite of behaviors to legged robotic systems introduces essential i... | [
-0.08329824358224869,
-0.0342290885746479,
0.0021859263069927692,
0.039374638348817825,
-0.014923958107829094,
-0.009880946949124336,
-0.007275545969605446,
0.01154952123761177,
-0.01415026094764471,
-0.005989158526062965,
-0.025765033438801765,
-0.02830052189528942,
-0.008380161598324776,
... | ||
corl_2024_9aZ4ehSTRc | 9aZ4ehSTRc | corl | 2,024 | Guided Reinforcement Learning for Robust Multi-Contact Loco-Manipulation | Reinforcement learning (RL) has shown remarkable proficiency in developing robust control policies for contact-rich applications. However, it typically requires meticulous Markov Decision Process (MDP) designing tailored to each task and robotic platform. This work addresses this challenge by creating a systematic appr... | Jean Pierre Sleiman;Mayank Mittal;Marco Hutter | ETHZ - ETH Zurich;NVIDIA;ETHZ - ETH Zurich | Poster | main | Whole-body Loco-Manipulation;Reinforcement Learning;Legged Mobile Manipulators | https://openreview.net/forum?id=9aZ4ehSTRc | 5 | Guided Reinforcement Learning for Robust Multi-Contact Loco-Manipulation
Reinforcement learning (RL) has shown remarkable proficiency in developing robust control policies for contact-rich applications. However, it typically requires meticulous Markov Decision Process (MDP) designing tailored to each task and robotic p... | [
-0.03437739983201027,
-0.03488831967115402,
-0.0015270481817424297,
0.004949506372213364,
-0.035070788115262985,
-0.012462720274925232,
0.016020890325307846,
0.01626722514629364,
0.015938779339194298,
-0.010327817872166634,
-0.05287988856434822,
-0.0206373892724514,
-0.007189329247921705,
... | ||
corl_2024_9dsBQhoqVr | 9dsBQhoqVr | corl | 2,024 | Fleet Supervisor Allocation: A Submodular Maximization Approach | In real-world scenarios, the data collected by robots in diverse and unpredictable environments is crucial for enhancing their perception and decision-making models. This data is predominantly collected under human supervision, particularly through imitation learning (IL), where robots learn complex tasks by observing ... | Oguzhan Akcin;Ahmet Ege Tanriverdi;Kaan Kale;Sandeep P. Chinchali | The University of Texas at Austin;Bogazici University;Bogazici University;University of Texas at Austin | Poster | main | Imitation Learning;Submodular Maximization;Fleet Learning | https://github.com/UTAustin-SwarmLab/Fleet-Supervisor-Allocation | https://openreview.net/forum?id=9dsBQhoqVr | 0 | Fleet Supervisor Allocation: A Submodular Maximization Approach
In real-world scenarios, the data collected by robots in diverse and unpredictable environments is crucial for enhancing their perception and decision-making models. This data is predominantly collected under human supervision, particularly through imitati... | [
-0.04397568851709366,
-0.05126776546239853,
-0.03571629524230957,
0.032553914934396744,
-0.03560468181967735,
-0.025224633514881134,
0.006180593278259039,
0.01975557580590248,
0.050970129668712616,
0.0551370307803154,
-0.01766282320022583,
0.0009952196851372719,
0.026452381163835526,
0.047... | |
corl_2024_9iG3SEbMnL | 9iG3SEbMnL | corl | 2,024 | ReKep: Spatio-Temporal Reasoning of Relational Keypoint Constraints for Robotic Manipulation | Representing robotic manipulation tasks as constraints that associate the robot and the environment is a promising way to encode desired robot behaviors. However, it remains unclear how to formulate the constraints such that they are 1) versatile to diverse tasks, 2) free of manual labeling, and 3) optimizable by off-t... | Wenlong Huang;Chen Wang;Yunzhu Li;Ruohan Zhang;Li Fei-Fei | NVIDIA;Computer Science Department, Stanford University;University of Illinois Urbana-Champaign;Stanford University;Stanford University | Poster | main | Structural Representation;Model-Based Planning;Foundation Models | https://github.com/huangwl18/ReKep | https://openreview.net/forum?id=9iG3SEbMnL | 97 | ReKep: Spatio-Temporal Reasoning of Relational Keypoint Constraints for Robotic Manipulation
Representing robotic manipulation tasks as constraints that associate the robot and the environment is a promising way to encode desired robot behaviors. However, it remains unclear how to formulate the constraints such that th... | [
-0.03746436536312103,
0.0018312942702323198,
-0.005376403219997883,
-0.0023323087953031063,
-0.013194532133638859,
0.046070296317338943,
0.022021600976586342,
0.02087905816733837,
0.009969611652195454,
0.031014522537589073,
-0.04146326705813408,
0.006260952912271023,
0.0026835945900529623,
... | |
corl_2024_9jJP2J1oBP | 9jJP2J1oBP | corl | 2,024 | Leveraging Mutual Information for Asymmetric Learning under Partial Observability | Even though partial observability is prevalent in robotics, most reinforcement learning studies avoid it due to the difficulty of learning a policy that can efficiently memorize past events and seek information. Fortunately, in many cases, learning can be done in an asymmetric setting where states are available during ... | Hai Huu Nguyen;Long Dinh Van The;Christopher Amato;Robert Platt | Northeastern University;Hanoi University of Science and Technology;Northeastern University;Northeastern University | Poster | main | Partial Observability;Mutual Information;Reinforcement Learning | https://sites.google.com/view/mi-asym-pomdp | https://openreview.net/forum?id=9jJP2J1oBP | 0 | Leveraging Mutual Information for Asymmetric Learning under Partial Observability
Even though partial observability is prevalent in robotics, most reinforcement learning studies avoid it due to the difficulty of learning a policy that can efficiently memorize past events and seek information. Fortunately, in many cases... | [
-0.03441100940108299,
0.00043985957745462656,
-0.007730146870017052,
0.005041194148361683,
-0.02213762328028679,
-0.040405429899692535,
0.029232285916805267,
0.04825888201594353,
0.02433810755610466,
-0.002541937865316868,
-0.053949784487485886,
-0.0424920953810215,
-0.004052399192005396,
... | |
corl_2024_A1hpY5RNiH | A1hpY5RNiH | corl | 2,024 | What Makes Pre-Trained Visual Representations Successful for Robust Manipulation? | Inspired by the success of transfer learning in computer vision, roboticists have investigated visual pre-training as a means to improve the learning efficiency and generalization ability of policies learned from pixels. To that end, past work has favored large object interaction datasets, such as first-person videos o... | Kaylee Burns;Zach Witzel;Jubayer Ibn Hamid;Tianhe Yu;Chelsea Finn;Karol Hausman | Stanford University;Stanford University;;Google Brain;Google; | Poster | main | representation learning;manipulation;visual features | https://openreview.net/forum?id=A1hpY5RNiH | 19 | What Makes Pre-Trained Visual Representations Successful for Robust Manipulation?
Inspired by the success of transfer learning in computer vision, roboticists have investigated visual pre-training as a means to improve the learning efficiency and generalization ability of policies learned from pixels. To that end, past... | [
-0.0523376390337944,
-0.029005952179431915,
-0.01887066848576069,
0.04072779044508934,
-0.021446485072374344,
-0.007251488044857979,
-0.012739100493490696,
0.021502481773495674,
0.03466155380010605,
0.026094157248735428,
-0.016238853335380554,
-0.024115631356835365,
-0.002008858136832714,
... | ||
corl_2024_A6ikGJRaKL | A6ikGJRaKL | corl | 2,024 | KOROL: Learning Visualizable Object Feature with Koopman Operator Rollout for Manipulation | Learning dexterous manipulation skills presents significant challenges due to complex nonlinear dynamics that underlie the interactions between objects and multi-fingered hands. Koopman operators have emerged as a robust method for modeling such nonlinear dynamics within a linear framework.
However, current methods rel... | Hongyi Chen;ABULIKEMU ABUDUWEILI;Aviral Agrawal;Yunhai Han;Harish Ravichandar;Changliu Liu;Jeffrey Ichnowski | Carnegie Mellon University;Carnegie Mellon University;Carnegie Mellon University;Georgia Institute of Technology;Georgia Institute of Technology;Carnegie Mellon University;Carnegie Mellon University | Poster | main | Manipulation;Koopman Operator;Visual Representation Learning | https://github.com/hychen-naza/KOROL | https://openreview.net/forum?id=A6ikGJRaKL | 5 | KOROL: Learning Visualizable Object Feature with Koopman Operator Rollout for Manipulation
Learning dexterous manipulation skills presents significant challenges due to complex nonlinear dynamics that underlie the interactions between objects and multi-fingered hands. Koopman operators have emerged as a robust method f... | [
-0.03883194550871849,
-0.029328925535082817,
0.0050123766995966434,
0.00584156159311533,
-0.02217371203005314,
0.01530731376260519,
0.019639572128653526,
0.017757602035999298,
0.017273135483264923,
0.00316068553365767,
-0.033987268805503845,
-0.03868287801742554,
0.012549575418233871,
0.03... | |
corl_2024_AEq0onGrN2 | AEq0onGrN2 | corl | 2,024 | Physically Embodied Gaussian Splatting: A Visually Learnt and Physically Grounded 3D Representation for Robotics | For robots to robustly understand and interact with the physical world, it is highly beneficial to have a comprehensive representation -- modelling geometry, physics, and visual observations -- that informs perception, planning, and control algorithms. We propose a novel dual "Gaussian-Particle" representation that mo... | Jad Abou-Chakra;Krishan Rana;Feras Dayoub;Niko Suenderhauf | Queensland University of Technology;Queensland University of Technology;University of Adelaide;Queensland University of Technology | Poster | main | 3D Representation;Gaussian Splatting;Robotics;Tracking;Physics | https://openreview.net/forum?id=AEq0onGrN2 | 9 | Physically Embodied Gaussian Splatting: A Visually Learnt and Physically Grounded 3D Representation for Robotics
For robots to robustly understand and interact with the physical world, it is highly beneficial to have a comprehensive representation -- modelling geometry, physics, and visual observations -- that informs... | [
-0.06009404733777046,
-0.03272244334220886,
0.0038845038507133722,
0.0036623317282646894,
-0.023236863315105438,
-0.052647773176431656,
0.019869202747941017,
-0.01624896563589573,
0.02566906437277794,
0.056950896978378296,
-0.01827891729772091,
0.017193783074617386,
0.05032783001661301,
0.... | ||
corl_2024_AGG1zlrrMw | AGG1zlrrMw | corl | 2,024 | Neural Attention Field: Emerging Point Relevance in 3D Scenes for One-Shot Dexterous Grasping | One-shot transfer of dexterous grasps to novel scenes with object and context variations has been a challenging problem. While distilled feature fields from large vision models have enabled semantic correspondences across 3D scenes, their features are point-based and restricted to object surfaces, limiting their capabi... | Qianxu Wang;Congyue Deng;Tyler Ga Wei Lum;Yuanpei Chen;Yaodong Yang;Jeannette Bohg;Yixin Zhu;Leonidas Guibas | Peking University;Stanford University;Stanford University;PsiRobot;Peking University;Stanford University;Peking University;Stanford University | Poster | main | Desterous Grasping;One-Shot Manipulation;Distilled Feature Field;Neural Implicit Field;Self-Supervised Learning | https://openreview.net/forum?id=AGG1zlrrMw | 2 | Neural Attention Field: Emerging Point Relevance in 3D Scenes for One-Shot Dexterous Grasping
One-shot transfer of dexterous grasps to novel scenes with object and context variations has been a challenging problem. While distilled feature fields from large vision models have enabled semantic correspondences across 3D s... | [
-0.0241929329931736,
-0.03498559072613716,
-0.04251872003078461,
-0.019466619938611984,
-0.009497899562120438,
-0.032939329743385315,
0.0005743784713558853,
0.0004122509271837771,
0.021911265328526497,
-0.030567120760679245,
-0.018162809312343597,
-0.0026913732290267944,
0.011870110407471657... | ||
corl_2024_AhEE5wrcLU | AhEE5wrcLU | corl | 2,024 | Velociraptor: Leveraging Visual Foundation Models for Label-Free, Risk-Aware Off-Road Navigation | Traversability analysis in off-road regimes is a challenging task that requires understanding of multi-modal inputs such as camera and LiDAR. These measurements are often sparse, noisy, and difficult to interpret, particularly in the off-road setting. Existing systems are very engineering-intensive, often requiring han... | Samuel Triest;Matthew Sivaprakasam;Shubhra Aich;David Fan;Wenshan Wang;Sebastian Scherer | Carnegie Mellon University;Carnegie Mellon University;Carnegie Mellon University;Jet Propulsion Laboratory;School of Computer Science, Carnegie Mellon University;Near Earth Autonomy Inc. | Poster | main | Field Robotics;Self-Supervised Learning;Visual Foundation Models | https://openreview.net/forum?id=AhEE5wrcLU | 2 | Velociraptor: Leveraging Visual Foundation Models for Label-Free, Risk-Aware Off-Road Navigation
Traversability analysis in off-road regimes is a challenging task that requires understanding of multi-modal inputs such as camera and LiDAR. These measurements are often sparse, noisy, and difficult to interpret, particula... | [
-0.07319066673517227,
-0.051326584070920944,
-0.01769240014255047,
0.019610628485679626,
-0.013436912558972836,
0.0037386836484074593,
0.03866254910826683,
-0.008152471855282784,
0.01418185606598854,
0.020746666938066483,
-0.030617164447903633,
-0.017841389402747154,
0.04261074960231781,
0... | ||
corl_2024_AsbyZRdqPv | AsbyZRdqPv | corl | 2,024 | Simple Masked Training Strategies Yield Control Policies That Are Robust to Sensor Failure | Sensor failure is common when robots are deployed in the real world, as sensors naturally wear out over time. Such failures can lead to catastrophic outcomes, including damage to the robot from unexpected robot behaviors such as falling during walking. Previous work has tried to address this problem by recovering missi... | Skand Skand;Bikram Pandit;Chanho Kim;Li Fuxin;Stefan Lee | Oregon State University;Oregon State University;Oregon State University;; | Poster | main | Reinforcement Learning;Robustness;Sensorimotor Learning | https://openreview.net/forum?id=AsbyZRdqPv | 1 | Simple Masked Training Strategies Yield Control Policies That Are Robust to Sensor Failure
Sensor failure is common when robots are deployed in the real world, as sensors naturally wear out over time. Such failures can lead to catastrophic outcomes, including damage to the robot from unexpected robot behaviors such as ... | [
-0.07531550526618958,
0.0050316136330366135,
-0.01373191736638546,
-0.007295372895896435,
-0.022814959287643433,
-0.025410115718841553,
-0.004301142878830433,
0.08125261962413788,
-0.01194891519844532,
0.009941871277987957,
-0.021713418886065483,
0.003117920132353902,
0.011911574751138687,
... | ||
corl_2024_AuJnXGq3AL | AuJnXGq3AL | corl | 2,024 | Scaling Cross-Embodied Learning: One Policy for Manipulation, Navigation, Locomotion and Aviation | Modern machine learning systems rely on large datasets to attain broad generalization, and this often poses a challenge in robotic learning, where each robotic platform and task might have only a small dataset. By training a single policy across many different kinds of robots, a robotic learning method can leverage muc... | Ria Doshi;Homer Rich Walke;Oier Mees;Sudeep Dasari;Sergey Levine | University of California, Berkeley;University of California, Berkeley;Electrical Engineering & Computer Science Department, University of California, Berkeley;Google;Carnegie Mellon University | Poster | main | Imitation Learning;Cross-Embodiment | https://github.com/rail-berkeley/crossformer | https://openreview.net/forum?id=AuJnXGq3AL | 53 | Scaling Cross-Embodied Learning: One Policy for Manipulation, Navigation, Locomotion and Aviation
Modern machine learning systems rely on large datasets to attain broad generalization, and this often poses a challenge in robotic learning, where each robotic platform and task might have only a small dataset. By training... | [
-0.07338128238916397,
-0.039050642400979996,
-0.06327752768993378,
0.0031712513882666826,
-0.04041501507163048,
-0.01965438388288021,
-0.020244384184479713,
0.020631570369005203,
0.01913813315331936,
0.006324064917862415,
-0.013855786994099617,
-0.008753206580877304,
0.04174251854419708,
-... | |
corl_2024_AzP6kSEffm | AzP6kSEffm | corl | 2,024 | Dynamics-Guided Diffusion Model for Sensor-less Robot Manipulator Design | We present Dynamics-Guided Diffusion Model (DGDM), a data-driven framework for generating task-specific manipulator designs without task-specific training. Given object shapes and task specifications, DGDM generates sensor-less manipulator designs that can blindly manipulate objects towards desired motions and poses u... | Xiaomeng Xu;Huy Ha;Shuran Song | Stanford University;Columbia University;Stanford University | Poster | main | manipulator design;hardware optimization;diffusion model | https://github.com/real-stanford/dgdm | https://openreview.net/forum?id=AzP6kSEffm | 1 | Dynamics-Guided Diffusion Model for Sensor-less Robot Manipulator Design
We present Dynamics-Guided Diffusion Model (DGDM), a data-driven framework for generating task-specific manipulator designs without task-specific training. Given object shapes and task specifications, DGDM generates sensor-less manipulator design... | [
-0.0826910212635994,
-0.04431409016251564,
-0.009603174403309822,
-0.01783829927444458,
-0.0064468239434063435,
-0.0005485053407028317,
-0.01731969229876995,
-0.011570305563509464,
0.024499719962477684,
0.03556929901242256,
-0.03499704599380493,
-0.018634093925356865,
0.016380833461880684,
... | |
corl_2024_B2X57y37kC | B2X57y37kC | corl | 2,024 | Learning to Look: Seeking Information for Decision Making via Policy Factorization | Many robot manipulation tasks require active or interactive exploration behavior in order to be performed successfully. Such tasks are ubiquitous in embodied domains, where agents must actively search for the information necessary for each stage of a task, e.g., moving the head of the robot to find information relevant... | Shivin Dass;Jiaheng Hu;Ben Abbatematteo;Peter Stone;Roberto Martín-Martín | University of Texas at Austin;University of Texas, Austin;University of Texas at Austin;Brown University;University of Texas at Austin | Poster | main | Active Vision;Manipulation;Interactive Perception | https://openreview.net/forum?id=B2X57y37kC | 0 | Learning to Look: Seeking Information for Decision Making via Policy Factorization
Many robot manipulation tasks require active or interactive exploration behavior in order to be performed successfully. Such tasks are ubiquitous in embodied domains, where agents must actively search for the information necessary for ea... | [
-0.0359133705496788,
-0.04787207394838333,
-0.00766512006521225,
0.02892814390361309,
-0.04649365693330765,
-0.0027195808943361044,
-0.01887873373925686,
0.002312109339982271,
0.014370935969054699,
0.05986803397536278,
0.008186683990061283,
-0.021924292668700218,
-0.013160163536667824,
0.0... | ||
corl_2024_B45HRM4Wb4 | B45HRM4Wb4 | corl | 2,024 | ResPilot: Teleoperated Finger Gaiting via Gaussian Process Residual Learning | Dexterous robot hand teleoperation allows for long-range transfer of human manipulation expertise, and could simultaneously provide a way for humans to teach these skills to robots. However, current methods struggle to reproduce the functional workspace of the human hand, often limiting them to simple grasping tasks. W... | Patrick Naughton;Jinda Cui;Karankumar Patel;Soshi Iba | Honda Research Institution US;Honda Research Institution US;Honda Research Institution US;Honda R&D | Poster | main | Teleoperation;Dexterous Manipulation;Gaussian Process | https://openreview.net/forum?id=B45HRM4Wb4 | 2 | ResPilot: Teleoperated Finger Gaiting via Gaussian Process Residual Learning
Dexterous robot hand teleoperation allows for long-range transfer of human manipulation expertise, and could simultaneously provide a way for humans to teach these skills to robots. However, current methods struggle to reproduce the functional... | [
-0.07066745311021805,
-0.002491601277142763,
-0.003368362784385681,
0.009068488143384457,
-0.01652918942272663,
-0.02181326411664486,
-0.012683660723268986,
0.017930127680301666,
-0.004294486250728369,
0.04565741866827011,
-0.055849477648735046,
0.005857613403350115,
0.024727966636419296,
... | ||
corl_2024_B7Lf6xEv7l | B7Lf6xEv7l | corl | 2,024 | DiffusionSeeder: Seeding Motion Optimization with Diffusion for Rapid Motion Planning | Running optimization across many parallel seeds leveraging GPU compute [2] have relaxed the need for a good initialization, but this can fail if the problem is highly non-convex as all seeds could get stuck in local minima. One such setting is collision-free motion optimization for robot manipulation, where optimizatio... | Huang Huang;Balakumar Sundaralingam;Arsalan Mousavian;Adithyavairavan Murali;Ken Goldberg;Dieter Fox | University of California, Berkeley;NVIDIA;NVIDIA;;University of California, Berkeley;Department of Computer Science | Poster | main | Robot Motion Planning;Diffusion Model | https://openreview.net/forum?id=B7Lf6xEv7l | 4 | DiffusionSeeder: Seeding Motion Optimization with Diffusion for Rapid Motion Planning
Running optimization across many parallel seeds leveraging GPU compute [2] have relaxed the need for a good initialization, but this can fail if the problem is highly non-convex as all seeds could get stuck in local minima. One such s... | [
-0.06357908993959427,
-0.03325091674923897,
-0.04167751967906952,
0.0271587111055851,
-0.0345035195350647,
0.04065266251564026,
-0.01577141135931015,
0.020022669807076454,
0.039893507957458496,
0.05382397025823593,
-0.026646282523870468,
0.004675909876823425,
-0.007064870558679104,
0.01338... | ||
corl_2024_BmvUg1FIWC | BmvUg1FIWC | corl | 2,024 | Neural Inverse Source Problem | Reconstructing unknown external source functions is an important perception capability for a large range of robotics domains including manipulation, aerial, and underwater robotics. In this work, we propose a Physics-Informed Neural Network (PINN) based approach for solving the inverse source problems in robotics, join... | Youngsun Wi;Jayjun Lee;Miquel Oller;Nima Fazeli | University of Michigan;;University of Michigan - Ann Arbor;University of Michigan | Poster | main | Inverse source problem;Physics informed neural network | https://openreview.net/forum?id=BmvUg1FIWC | 2 | Neural Inverse Source Problem
Reconstructing unknown external source functions is an important perception capability for a large range of robotics domains including manipulation, aerial, and underwater robotics. In this work, we propose a Physics-Informed Neural Network (PINN) based approach for solving the inverse sou... | [
-0.04130327329039574,
-0.016444474458694458,
-0.01121597457677126,
0.021213840693235397,
-0.0028719266410917044,
-0.004839641507714987,
0.028691157698631287,
0.009590267203748226,
0.030583912506699562,
0.052959639579057693,
-0.056820109486579895,
0.0342007577419281,
-0.01624770276248455,
0... | ||
corl_2024_Bq4XOaU4sV | Bq4XOaU4sV | corl | 2,024 | Bridging the Sim-to-Real Gap from the Information Bottleneck Perspective | Reinforcement Learning (RL) has recently achieved remarkable success in robotic control. However, most works in RL operate in simulated environments where privileged knowledge (e.g., dynamics, surroundings, terrains) is readily available. Conversely, in real-world scenarios, robot agents usually rely solely on local st... | Haoran He;Peilin Wu;Chenjia Bai;Hang Lai;Lingxiao Wang;Ling Pan;Xiaolin Hu;Weinan Zhang | Hong Kong University of Science and Technology;Shanghai Jiaotong University;Shanghai AI Laboratory;Shanghai Jiaotong University;Northwestern University;Montreal Institute for Learning Algorithms (MILA);Tsinghua University;Shanghai Jiaotong University | Poster | main | Sim-to-Real;Information Bottleneck;Reinforcement Learning;Locomotion | https://github.com/tinnerhrhe/HIB_Policy | https://openreview.net/forum?id=Bq4XOaU4sV | 9 | Bridging the Sim-to-Real Gap from the Information Bottleneck Perspective
Reinforcement Learning (RL) has recently achieved remarkable success in robotic control. However, most works in RL operate in simulated environments where privileged knowledge (e.g., dynamics, surroundings, terrains) is readily available. Converse... | [
-0.06423180550336838,
-0.04795466363430023,
0.03560385853052139,
0.004638698883354664,
-0.006142047233879566,
-0.0549686998128891,
0.02008911594748497,
0.03802446648478508,
-0.0005741789354942739,
0.021251767873764038,
-0.033793170005083084,
-0.022757498547434807,
-0.04383772984147072,
0.0... | |
corl_2024_CPQW5kc0pe | CPQW5kc0pe | corl | 2,024 | VoxAct-B: Voxel-Based Acting and Stabilizing Policy for Bimanual Manipulation | Bimanual manipulation is critical to many robotics applications. In contrast to single-arm manipulation, bimanual manipulation tasks are challenging due to higher-dimensional action spaces. Prior works leverage large amounts of data and primitive actions to address this problem, but may suffer from sample inefficiency ... | I-Chun Arthur Liu;Sicheng He;Daniel Seita;Gaurav S. Sukhatme | University of Southern California;;;University of Southern California | Poster | main | bimanual manipulation;voxel representation;vision language models | https://github.com/VoxAct-B/voxactb | https://openreview.net/forum?id=CPQW5kc0pe | 14 | VoxAct-B: Voxel-Based Acting and Stabilizing Policy for Bimanual Manipulation
Bimanual manipulation is critical to many robotics applications. In contrast to single-arm manipulation, bimanual manipulation tasks are challenging due to higher-dimensional action spaces. Prior works leverage large amounts of data and primi... | [
-0.035452429205179214,
-0.004802964627742767,
-0.017716987058520317,
0.012235794216394424,
-0.02127883955836296,
0.01930413395166397,
0.029399126768112183,
0.048168059438467026,
0.00801878236234188,
0.019931610673666,
-0.024028664454817772,
-0.034880317747592926,
0.02116810902953148,
-0.01... | |
corl_2024_CpXiqz6qf4 | CpXiqz6qf4 | corl | 2,024 | SonicSense: Object Perception from In-Hand Acoustic Vibration | We introduce SonicSense, a holistic design of hardware and software to enable rich robot object perception through in-hand acoustic vibration sensing. While previous studies have shown promising results with acoustic sensing for object perception, current solutions are constrained to a handful of objects with simple ge... | Jiaxun Liu;Boyuan Chen | Duke University;Duke University | Poster | main | Tactile Perception;Object State Estimation;Audio;Acoustic Vibration Sensing | https://github.com/generalroboticslab/SonicSense?tab=readme-ov-file | https://openreview.net/forum?id=CpXiqz6qf4 | 5 | SonicSense: Object Perception from In-Hand Acoustic Vibration
We introduce SonicSense, a holistic design of hardware and software to enable rich robot object perception through in-hand acoustic vibration sensing. While previous studies have shown promising results with acoustic sensing for object perception, current so... | [
-0.038686275482177734,
0.005439676810055971,
-0.005802012048661709,
0.015626877546310425,
0.015599004924297333,
-0.03998696804046631,
0.007678723894059658,
0.0013750165235251188,
0.04058156907558441,
0.03344634920358658,
0.016908986493945122,
0.009606534615159035,
0.016072828322649002,
-0.... | |
corl_2024_CskuWHDBAr | CskuWHDBAr | corl | 2,024 | Enhancing Visual Domain Robustness in Behaviour Cloning via Saliency-Guided Augmentation | In vision-based behaviour cloning (BC), traditional image-level augmentation methods such as pixel shifting enhance in-domain performance but often struggle with visual domain shifts, including distractors, occlusion, and changes in lighting and backgrounds. Conversely, superimposition-based augmentation, proven effect... | Zheyu Zhuang;RUIYU WANG;Nils Ingelhag;Ville Kyrki;Danica Kragic | KTH Royal Institute of Technology;KTH Royal Institute of Technology;;Aalto University;KTH | Poster | main | Behaviour Cloning;Visuomotor Policy;Data Augmentation | https://github.com/Zheyu-Zhuang/RoboSaGA | https://openreview.net/forum?id=CskuWHDBAr | 2 | Enhancing Visual Domain Robustness in Behaviour Cloning via Saliency-Guided Augmentation
In vision-based behaviour cloning (BC), traditional image-level augmentation methods such as pixel shifting enhance in-domain performance but often struggle with visual domain shifts, including distractors, occlusion, and changes i... | [
-0.06226715072989464,
-0.014366479590535164,
-0.01053293701261282,
0.0041824690997600555,
-0.01064459327608347,
-0.017353292554616928,
0.042690031230449677,
0.04544422775506973,
0.03260372206568718,
0.006908750627189875,
-0.021084483712911606,
0.011268009431660175,
-0.021903298795223236,
0... | |
corl_2024_Czs2xH9114 | Czs2xH9114 | corl | 2,024 | WoCoCo: Learning Whole-Body Humanoid Control with Sequential Contacts | Humanoid activities involving sequential contacts are crucial for complex robotic interactions and operations in the real world and are traditionally solved by model-based motion planning, which is time-consuming and often relies on simplified dynamics models.
Although model-free reinforcement learning (RL) has become... | Chong Zhang;Wenli Xiao;Tairan He;Guanya Shi | ETHZ - ETH Zurich;Carnegie Mellon University;Carnegie Mellon University;Carnegie Mellon University | Poster | main | Whole-Body Humanoid Control;Multi-Contact Control;Reinforcement Learning | https://openreview.net/forum?id=Czs2xH9114 | 45 | WoCoCo: Learning Whole-Body Humanoid Control with Sequential Contacts
Humanoid activities involving sequential contacts are crucial for complex robotic interactions and operations in the real world and are traditionally solved by model-based motion planning, which is time-consuming and often relies on simplified dynami... | [
-0.046264369040727615,
0.004479791969060898,
-0.018066272139549255,
-0.008559643290936947,
-0.0087389275431633,
-0.007690807804465294,
-0.0030822991393506527,
-0.020557856187224388,
-0.006987464148551226,
-0.0026547766756266356,
-0.009331942535936832,
-0.028354395180940628,
0.018627109006047... | ||
corl_2024_DDIoRSh8ID | DDIoRSh8ID | corl | 2,024 | Multi-Task Interactive Robot Fleet Learning with Visual World Models | Recent advancements in large-scale multi-task robot learning offer the potential for deploying robot fleets in household and industrial settings, enabling them to perform diverse tasks across various environments. However, AI-enabled robots often face challenges with generalization and robustness when exposed to real-w... | Huihan Liu;Yu Zhang;Vaarij Betala;Evan Zhang;James Liu;Crystal Ding;Yuke Zhu | ;ShanghaiTech University;;;;;Computer Science Department, University of Texas, Austin | Poster | main | Robot Manipulation;Interactive Imitation Learning;Fleet Learning | https://openreview.net/forum?id=DDIoRSh8ID | 4 | Multi-Task Interactive Robot Fleet Learning with Visual World Models
Recent advancements in large-scale multi-task robot learning offer the potential for deploying robot fleets in household and industrial settings, enabling them to perform diverse tasks across various environments. However, AI-enabled robots often face... | [
-0.0532115176320076,
-0.04185471311211586,
-0.01719502918422222,
0.038808684796094894,
-0.002481950679793954,
-0.020062435418367386,
-0.011046561412513256,
0.016762569546699524,
0.031231213361024857,
0.04632974788546562,
0.010802126489579678,
-0.003903901670128107,
-0.008019332773983479,
0... | ||
corl_2024_DSdAEsEGhE | DSdAEsEGhE | corl | 2,024 | SoloParkour: Constrained Reinforcement Learning for Visual Locomotion from Privileged Experience | Parkour poses a significant challenge for legged robots, requiring navigation through complex environments with agility and precision based on limited sensory inputs.
In this work, we introduce a novel method for training end-to-end visual policies, from depth pixels to robot control commands, to achieve agile and saf... | Elliot Chane-Sane;Joseph Amigo;Thomas Flayols;Ludovic Righetti;Nicolas Mansard | LAAS / CNRS;New York University;;Max-Planck Institute;LAAS / CNRS | Poster | main | Reinforcement Learning;Agile Locomotion;Visuomotor Control | https://github.com/Gepetto/SoloParkour | https://openreview.net/forum?id=DSdAEsEGhE | 7 | SoloParkour: Constrained Reinforcement Learning for Visual Locomotion from Privileged Experience
Parkour poses a significant challenge for legged robots, requiring navigation through complex environments with agility and precision based on limited sensory inputs.
In this work, we introduce a novel method for training ... | [
-0.0784868374466896,
-0.03239520266652107,
0.00666824309155345,
0.027374407276511192,
-0.051869239658117294,
0.019141780212521553,
-0.00876793172210455,
0.004503948148339987,
0.011896699666976929,
0.003883271012455225,
-0.012644280679523945,
-0.01934482716023922,
0.01447170227766037,
0.015... | |
corl_2024_Dftu4r5jHe | Dftu4r5jHe | corl | 2,024 | Context-Aware Replanning with Pre-Explored Semantic Map for Object Navigation | Pre-explored Semantic Map, constructed through prior exploration using visual language models (VLMs), has proven effective as a foundational element for training-free robotic applications. However, existing approaches assume the map's accuracy and do not provide effective mechanisms for revising decisions based on inco... | Po-Chen Ko;Hung-Ting Su;CY Chen;Jia-Fong Yeh;Min Sun;Winston H. Hsu | National Taiwan University;National Taiwan University;Sony Group Corporation;National Tsing Hua University;National Taiwan University;National Taiwan University | Poster | main | VLMs;map;navigation;uncertainty;multi-view consistency;robotics | https://github.com/CARe-maps/CARe_experiments | https://openreview.net/forum?id=Dftu4r5jHe | 0 | Context-Aware Replanning with Pre-Explored Semantic Map for Object Navigation
Pre-explored Semantic Map, constructed through prior exploration using visual language models (VLMs), has proven effective as a foundational element for training-free robotic applications. However, existing approaches assume the map's accurac... | [
-0.06104801222681999,
-0.011282220482826233,
-0.008098394609987736,
0.018075617030262947,
0.004113984294235706,
-0.0036072563380002975,
0.05012675002217293,
0.02167593128979206,
0.010967540554702282,
0.016798384487628937,
0.015697002410888672,
-0.012448389083147049,
-0.004294462967664003,
... | |
corl_2024_DsFQg0G4Xu | DsFQg0G4Xu | corl | 2,024 | Learning Long-Horizon Action Dependencies in Sampling-Based Bilevel Planning | Autonomous robots will need the ability to make task and motion plans that involve long sequences of actions, e.g. to prepare a meal. One challenge is that the feasibility of actions late in the plan may depend on much earlier actions. This issue is exacerbated if these dependencies exist at a purely geometric level, m... | Bartłomiej Cieślar;Leslie Pack Kaelbling;Tomás Lozano-Pérez;Jorge Mendez-Mendez | Imperial College London;Massachusetts Institute of Technology;Massachusetts Institute of Technology;Massachusetts Institute of Technology | Poster | main | task and motion planning;long-horizon;learning for planning | https://openreview.net/forum?id=DsFQg0G4Xu | 0 | Learning Long-Horizon Action Dependencies in Sampling-Based Bilevel Planning
Autonomous robots will need the ability to make task and motion plans that involve long sequences of actions, e.g. to prepare a meal. One challenge is that the feasibility of actions late in the plan may depend on much earlier actions. This is... | [
-0.02016174979507923,
-0.020947271957993507,
0.0025319079868495464,
0.022742751985788345,
-0.0075933863408863544,
0.012241062708199024,
0.006648888811469078,
0.00020748554379679263,
-0.037218812853097916,
0.008430342189967632,
-0.06119595840573311,
-0.01009022630751133,
0.00980968214571476,
... | ||
corl_2024_E4K3yLQQ7s | E4K3yLQQ7s | corl | 2,024 | Visual Manipulation with Legs | Animals have the ability to use their arms and legs for both locomotion and manipulation. We envision quadruped robots to have the same versatility. This work presents a system that empowers a quadruped robot to perform object interactions with its legs, drawing inspiration from non-prehensile manipulation techniques. ... | Xialin He;Chengjing Yuan;Wenxuan Zhou;Ruihan Yang;David Held;Xiaolong Wang | Shanghai Jiaotong University;University of California, San Diego;Carnegie Mellon University;University of California, San Diego;Carnegie Mellon University;University of California, San Diego | Poster | main | Legged robots;Non-prehensile manipulation;Reinforcement Learning | https://openreview.net/forum?id=E4K3yLQQ7s | 2 | Visual Manipulation with Legs
Animals have the ability to use their arms and legs for both locomotion and manipulation. We envision quadruped robots to have the same versatility. This work presents a system that empowers a quadruped robot to perform object interactions with its legs, drawing inspiration from non-prehen... | [
-0.052956998348236084,
-0.032059844583272934,
-0.02877117320895195,
0.06423243880271912,
-0.027061063796281815,
-0.005647117737680674,
0.0040474142879247665,
-0.004432658664882183,
0.0012449967907741666,
-0.00848477054387331,
-0.030049055814743042,
-0.008465978316962719,
0.025275785475969315... | ||
corl_2024_EM0wndCeoD | EM0wndCeoD | corl | 2,024 | BiGym: A Demo-Driven Mobile Bi-Manual Manipulation Benchmark | We introduce BiGym, a new benchmark and learning environment for mobile bi-manual demo-driven robotic manipulation. BiGym features 40 diverse tasks set in home environments, ranging from simple target reaching to complex kitchen cleaning. To capture the real-world performance accurately, we provide human-collected demo... | Nikita Chernyadev;Nicholas Backshall;Xiao Ma;Yunfan Lu;Younggyo Seo;Stephen James | Dyson;;Dyson Robot Learning Lab;;Dyson;Dyson | Poster | main | Bi-Manual Manipulation;Mobile Manipulation;Benchmark | https://github.com/chernyadev/bigym | https://openreview.net/forum?id=EM0wndCeoD | 10 | BiGym: A Demo-Driven Mobile Bi-Manual Manipulation Benchmark
We introduce BiGym, a new benchmark and learning environment for mobile bi-manual demo-driven robotic manipulation. BiGym features 40 diverse tasks set in home environments, ranging from simple target reaching to complex kitchen cleaning. To capture the real-... | [
-0.0676727220416069,
-0.05718345195055008,
-0.02511409856379032,
-0.028328552842140198,
-0.023234302178025246,
-0.0053057293407619,
-0.006104643456637859,
-0.010367085225880146,
-0.004715942777693272,
0.04680696502327919,
-0.01775469072163105,
0.018478412181138992,
0.011100206524133682,
0.... | |
corl_2024_EPujQZWemk | EPujQZWemk | corl | 2,024 | ViPER: Visibility-based Pursuit-Evasion via Reinforcement Learning | In visibility-based pursuit-evasion tasks, a team of mobile pursuer robots with limited sensing capabilities is tasked with detecting all evaders in a multiply-connected planar environment, whose map may or may not be known to pursuers beforehand. This requires tight coordination among multiple agents to ensure that th... | Yizhuo Wang;Yuhong Cao;Jimmy Chiun;Subhadeep Koley;Mandy Pham;Guillaume Adrien Sartoretti | National University of Singapore;National University of Singapore;;Indian Institute of Engineering Science and Technology, Shibpur;University of California, Berkeley;National University of Singapore | Poster | main | MARL;pursuit-evasion;graph attention;path planning | https://github.com/marmotlab/ViPER | https://openreview.net/forum?id=EPujQZWemk | 1 | ViPER: Visibility-based Pursuit-Evasion via Reinforcement Learning
In visibility-based pursuit-evasion tasks, a team of mobile pursuer robots with limited sensing capabilities is tasked with detecting all evaders in a multiply-connected planar environment, whose map may or may not be known to pursuers beforehand. This ... | [
-0.10091979801654816,
-0.039164017885923386,
-0.0034324252046644688,
0.023836543783545494,
-0.04815613478422165,
0.010859282687306404,
0.020102214068174362,
0.01737113669514656,
0.042916927486658096,
0.005870886612683535,
-0.02246171608567238,
-0.026381833478808403,
-0.016191385686397552,
... | |
corl_2024_EdVNB2kHv1 | EdVNB2kHv1 | corl | 2,024 | Scaling Robot Policy Learning via Zero-Shot Labeling with Foundation Models | A central challenge towards developing robots that can relate human language to their perception and actions is the scarcity of natural language annotations in diverse robot datasets. Moreover, robot policies that follow natural language instructions are typically trained on either templated language or expensive human... | Nils Blank;Moritz Reuss;Marcel Rühle;Ömer Erdinç Yağmurlu;Fabian Wenzel;Oier Mees;Rudolf Lioutikov | Karlsruher Institut für Technologie;Karlsruher Institut für Technologie;Karlsruher Institut für Technologie;Karlsruher Institut für Technologie;Karlsruher Institut für Technologie;Electrical Engineering & Computer Science Department, University of California, Berkeley;Karlsruher Institut für Technologie | Poster | main | Foundation Models;Language-conditioned Imitation Learning;Data Labeling | https://openreview.net/forum?id=EdVNB2kHv1 | 7 | Scaling Robot Policy Learning via Zero-Shot Labeling with Foundation Models
A central challenge towards developing robots that can relate human language to their perception and actions is the scarcity of natural language annotations in diverse robot datasets. Moreover, robot policies that follow natural language instru... | [
-0.09158093482255936,
-0.01840776763856411,
-0.007440950721502304,
0.04674290865659714,
-0.008439183235168457,
-0.025129808112978935,
0.023298189043998718,
0.0006330531905405223,
0.01294038537889719,
0.030404869467020035,
-0.033390406519174576,
-0.007463845890015364,
0.016411302611231804,
... | ||
corl_2024_EifoVoIyd5 | EifoVoIyd5 | corl | 2,024 | What Matters in Range View 3D Object Detection | Lidar-based perception pipelines rely on 3D object detection models to interpret complex scenes. While multiple representations for lidar exist, the range view is enticing since it losslessly encodes the entire lidar sensor output. In this work, we achieve state-of-the-art amongst range view 3D object detection models ... | Benjamin Wilson;Nicholas Autio Mitchell;Jhony Kaesemodel Pontes;James Hays | Georgia Institute of Technology;NVIDIA;Latitude AI;Georgia Institute of Technology | Poster | main | 3D Object Detection;3D Perception;Autonomous Driving | https://github.com/benjaminrwilson/range-view-3d-detection | https://openreview.net/forum?id=EifoVoIyd5 | 2 | What Matters in Range View 3D Object Detection
Lidar-based perception pipelines rely on 3D object detection models to interpret complex scenes. While multiple representations for lidar exist, the range view is enticing since it losslessly encodes the entire lidar sensor output. In this work, we achieve state-of-the-art... | [
-0.04061802849173546,
-0.022916005924344063,
-0.048785384744405746,
-0.0012248759157955647,
0.01111161895096302,
-0.036826036870479584,
0.003281528130173683,
0.021767470985651016,
0.021147625520825386,
0.010901966132223606,
-0.03926895558834076,
-0.03606034815311432,
0.004354861564934254,
... | |
corl_2024_EiqQEsOMZt | EiqQEsOMZt | corl | 2,024 | TaMMa: Target-driven Multi-subscene Mobile Manipulation | For everyday service robotics, the ability to navigate back and forth based on tasks in multi-subscene environments and perform delicate manipulations is crucial and highly practical.
While existing robotics primarily focus on complex tasks within a single scene or simple tasks across scalable scenes individually, robo... | Jiawei Hou;Tianyu Wang;Tongying Pan;Shouyan Wang;Xiangyang Xue;Yanwei Fu | Fudan University;Fudan University;Fudan University;Fudan University;Fudan University;Fudan University, | Poster | main | Multi-subscene;3D Gaussians;Scene Inpainting;Target-driven Mobile Manipulation | https://openreview.net/forum?id=EiqQEsOMZt | 2 | TaMMa: Target-driven Multi-subscene Mobile Manipulation
For everyday service robotics, the ability to navigate back and forth based on tasks in multi-subscene environments and perform delicate manipulations is crucial and highly practical.
While existing robotics primarily focus on complex tasks within a single scene o... | [
-0.015687253326177597,
-0.0005512507050298154,
-0.029403137043118477,
-0.012293152511119843,
-0.03461052477359772,
0.02092253603041172,
-0.0013739131391048431,
-0.009754551574587822,
0.008452704176306725,
0.07167595624923706,
-0.0415475070476532,
0.025125641375780106,
0.021685045212507248,
... | ||
corl_2024_EyEE7547vy | EyEE7547vy | corl | 2,024 | Event3DGS: Event-Based 3D Gaussian Splatting for High-Speed Robot Egomotion | By combining differentiable rendering with explicit point-based scene representations, 3D Gaussian Splatting (3DGS) has demonstrated breakthrough 3D reconstruction capabilities.
However, to date 3DGS has had limited impact on robotics, where high-speed egomotion is pervasive: Egomotion introduces motion blur and l... | Tianyi Xiong;Jiayi Wu;Botao He;Cornelia Fermuller;Yiannis Aloimonos;Heng Huang;Christopher Metzler | University of Maryland, College Park;University of Maryland, College Park;University of Maryland, College Park;University of Maryland, College Park;University of Maryland, College Park;Department of Computer Science, University of Maryland, College Park;University of Maryland, College Park | Poster | main | Event-based 3D Reconstruction;Gaussian Splatting;High-speed Robot Egomotion | https://openreview.net/forum?id=EyEE7547vy | 11 | Event3DGS: Event-Based 3D Gaussian Splatting for High-Speed Robot Egomotion
By combining differentiable rendering with explicit point-based scene representations, 3D Gaussian Splatting (3DGS) has demonstrated breakthrough 3D reconstruction capabilities.
However, to date 3DGS has had limited impact on robotics, whe... | [
-0.05147513747215271,
-0.04223791882395744,
-0.010484986007213593,
-0.022999925538897514,
-0.03179017826914787,
-0.009186002425849438,
0.014507644809782505,
-0.01998293027281761,
-0.005251804832369089,
0.028531081974506378,
-0.007472647819668055,
0.036613646894693375,
0.02519748918712139,
... | ||
corl_2024_F0rWEID2gb | F0rWEID2gb | corl | 2,024 | Environment Curriculum Generation via Large Language Models | Recent work has demonstrated that a promising strategy for teaching robots a wide range of complex skills is by training them on a curriculum of progressively more challenging environments. However, developing an effective curriculum of environment distributions currently requires significant expertise, which must be r... | William Liang;Sam Wang;Hung-Ju Wang;Osbert Bastani;Dinesh Jayaraman;Yecheng Jason Ma | University of Pennsylvania;University of Pennsylvania;;University of Pennsylvania;University of Pennsylvania; | Poster | main | Large Language Models;Environment Curriculum;Quadrupeds;Sim-To-Real Reinforcement Learning | https://openreview.net/forum?id=F0rWEID2gb | 4 | Environment Curriculum Generation via Large Language Models
Recent work has demonstrated that a promising strategy for teaching robots a wide range of complex skills is by training them on a curriculum of progressively more challenging environments. However, developing an effective curriculum of environment distributio... | [
-0.06761855632066727,
0.010745727457106113,
0.0013582432875409722,
-0.02639269083738327,
-0.027114003896713257,
0.011217355728149414,
-0.030406154692173004,
-0.027927793562412262,
-0.0013836742145940661,
0.0003022813762072474,
0.00015417506801895797,
0.01618332415819168,
-0.01379743963479995... | ||
corl_2024_FHnVRmeqxf | FHnVRmeqxf | corl | 2,024 | FlowRetrieval: Flow-Guided Data Retrieval for Few-Shot Imitation Learning | Imitation learning policies in robotics tend to require an extensive amount of demonstrations. It is critical to develop few-shot adaptation strategies that rely only on a small amount of task-specific human demonstrations. Prior works focus on learning general policies from large scale dataset with diverse behaviors. ... | Li-Heng Lin;Yuchen Cui;Amber Xie;Tianyu Hua;Dorsa Sadigh | Stanford University;Stanford University;;;Stanford University | Poster | main | Data Retrieval;Few-shot Learning;Imitation Learning | https://openreview.net/forum?id=FHnVRmeqxf | 9 | FlowRetrieval: Flow-Guided Data Retrieval for Few-Shot Imitation Learning
Imitation learning policies in robotics tend to require an extensive amount of demonstrations. It is critical to develop few-shot adaptation strategies that rely only on a small amount of task-specific human demonstrations. Prior works focus on l... | [
-0.06562439352273941,
-0.005118375178426504,
-0.03705248609185219,
0.018107673153281212,
-0.03352194279432297,
-0.0009906899649649858,
0.013758191838860512,
0.022457154467701912,
0.003000505268573761,
0.011556153185665607,
-0.03890874981880188,
-0.03417709842324257,
-0.006515122950077057,
... | ||
corl_2024_FO6tePGRZj | FO6tePGRZj | corl | 2,024 | Mobile ALOHA: Learning Bimanual Mobile Manipulation using Low-Cost Whole-Body Teleoperation | Imitation learning from human demonstrations has shown impressive performance in robotics. However, most results focus on table-top manipulation, lacking the mobility and dexterity necessary for generally useful tasks. In this work, we develop a system for imitating mobile manipulation tasks that are bimanual and requi... | Zipeng Fu;Tony Z. Zhao;Chelsea Finn | Stanford University;Stanford University;Google | Poster | main | Mobile Manipulation;Imitation Learning | https://github.com/MarkFzp/mobile-aloha | https://openreview.net/forum?id=FO6tePGRZj | 15 | Mobile ALOHA: Learning Bimanual Mobile Manipulation using Low-Cost Whole-Body Teleoperation
Imitation learning from human demonstrations has shown impressive performance in robotics. However, most results focus on table-top manipulation, lacking the mobility and dexterity necessary for generally useful tasks. In this w... | [
-0.04515749216079712,
-0.02227894403040409,
-0.047368522733449936,
0.043358687311410904,
-0.02389037422835827,
0.002148962579667568,
0.00419018417596817,
0.041035231202840805,
-0.023628048598766327,
0.023478148505091667,
-0.03290313482284546,
-0.019262200221419334,
0.009120501577854156,
0.... | |
corl_2024_G0jqGG8Tta | G0jqGG8Tta | corl | 2,024 | Not All Errors Are Made Equal: A Regret Metric for Detecting System-level Trajectory Prediction Failures | Robot decision-making increasingly relies on data-driven human prediction models when operating around people. While these models are known to mispredict in out-of-distribution interactions, only a subset of prediction errors impact downstream robot performance.
We propose characterizing such ``system-level'' predict... | Kensuke Nakamura;Thomas Tian;Andrea Bajcsy | Carnegie Mellon University;University of California, Berkeley;Carnegie Mellon University | Poster | main | Human-Robot Interaction;Trajectory Prediction;Failure Detection | https://openreview.net/forum?id=G0jqGG8Tta | 1 | Not All Errors Are Made Equal: A Regret Metric for Detecting System-level Trajectory Prediction Failures
Robot decision-making increasingly relies on data-driven human prediction models when operating around people. While these models are known to mispredict in out-of-distribution interactions, only a subset of predict... | [
-0.043198224157094955,
-0.052999190986156464,
-0.03446981683373451,
0.013767584227025509,
-0.022153884172439575,
-0.0220614206045866,
0.013462459668517113,
-0.005737264174968004,
-0.022893579676747322,
0.02664753422141075,
-0.05336903780698776,
-0.019028669223189354,
-0.011261865496635437,
... | ||
corl_2024_G8UcwxNAoD | G8UcwxNAoD | corl | 2,024 | Teaching Robots with Show and Tell: Using Foundation Models to Synthesize Robot Policies from Language and Visual Demonstration | We introduce a modular, neuro-symbolic framework for teaching robots new skills through language and visual demonstration. Our approach, ShowTell, composes a mixture of foundation models to synthesize robot manipulation programs that are easy to interpret and generalize across a wide range of tasks and environments. Sh... | Michael Murray;Abhishek Gupta;Maya Cakmak | University of Washington;University of Washington;University of Washington, Seattle | Poster | main | learning from demonstration;language model planning;neuro-symbolic reasoning | https://openreview.net/forum?id=G8UcwxNAoD | 2 | Teaching Robots with Show and Tell: Using Foundation Models to Synthesize Robot Policies from Language and Visual Demonstration
We introduce a modular, neuro-symbolic framework for teaching robots new skills through language and visual demonstration. Our approach, ShowTell, composes a mixture of foundation models to sy... | [
-0.02947467379271984,
-0.02433808520436287,
0.010600049048662186,
0.004034555982798338,
-0.019519032910466194,
-0.013485876843333244,
0.0031309835612773895,
0.027662858366966248,
0.004429139196872711,
0.0071725440211594105,
-0.04127948358654976,
0.007279945537447929,
0.006247958168387413,
... | ||
corl_2024_GGuNkjQSrk | GGuNkjQSrk | corl | 2,024 | Action Space Design in Reinforcement Learning for Robot Motor Skills | Practitioners often rely on intuition to select action spaces for learning. The choice can substantially impact final performance even when choosing among configuration-space representations such as joint position, velocity, and torque commands. We examine action space selection considering a wheeled-legged robot, a qu... | Julian Eßer;Gabriel B. Margolis;Oliver Urbann;Sören Kerner;Pulkit Agrawal | Fraunhofer IML;;Fraunhofer IML;;Massachusetts Institute of Technology | Poster | main | Reinforcement Learning;Action Spaces;Sim-to-Real | https://openreview.net/forum?id=GGuNkjQSrk | 1 | Action Space Design in Reinforcement Learning for Robot Motor Skills
Practitioners often rely on intuition to select action spaces for learning. The choice can substantially impact final performance even when choosing among configuration-space representations such as joint position, velocity, and torque commands. We ex... | [
-0.0536654070019722,
-0.02753252536058426,
-0.017016759142279625,
0.014650250785052776,
-0.032615453004837036,
-0.005593982990831137,
0.004940200597047806,
0.003611917607486248,
0.05138177052140236,
0.010303978808224201,
-0.00026588686159811914,
-0.015396115370094776,
-0.020534291863441467,
... | ||
corl_2024_GVX6jpZOhU | GVX6jpZOhU | corl | 2,024 | RoboPoint: A Vision-Language Model for Spatial Affordance Prediction in Robotics | From rearranging objects on a table to putting groceries into shelves, robots must plan precise action points to perform tasks accurately and reliably. In spite of the recent adoption of vision language models (VLMs) to control robot behavior, VLMs struggle to precisely articulate robot actions using language. We intro... | Wentao Yuan;Jiafei Duan;Valts Blukis;Wilbert Pumacay;Ranjay Krishna;Adithyavairavan Murali;Arsalan Mousavian;Dieter Fox | University of Washington, Seattle;NVIDIA;NVIDIA;Universidad Nacional de Ingeniería;University of Washington;;NVIDIA;Department of Computer Science | Poster | main | Foundation Model;Affordance Prediction;Open-world Manipulation | https://github.com/wentaoyuan/RoboPoint | https://openreview.net/forum?id=GVX6jpZOhU | 53 | RoboPoint: A Vision-Language Model for Spatial Affordance Prediction in Robotics
From rearranging objects on a table to putting groceries into shelves, robots must plan precise action points to perform tasks accurately and reliably. In spite of the recent adoption of vision language models (VLMs) to control robot behav... | [
-0.03870873153209686,
-0.026715250685811043,
0.010382143780589104,
0.018127551302313805,
-0.029828062281012535,
0.0010414187563583255,
0.00782322883605957,
0.0023689414374530315,
0.001032263389788568,
0.0096130957826972,
-0.03420431166887283,
-0.0045547764748334885,
0.034405726939439774,
0... | |
corl_2024_HlxRd529nG | HlxRd529nG | corl | 2,024 | Detect Everything with Few Examples | Few-shot object detection aims at detecting novel categories given only a few example images. It is a basic skill for a robot to perform tasks in open environments. Recent methods focus on finetuning strategies, with complicated procedures that prohibit a wider application. In this paper, we introduce DE-ViT, a few-sho... | Xinyu Zhang;Yuhan Liu;Yuting Wang;Abdeslam Boularias | Rutgers University;Rutgers University;Amazon;, Rutgers University | Poster | main | Robot Vision;Object Detection and Recognition;Few-shot Learning | http://github.com/mlzxy/devit | https://openreview.net/forum?id=HlxRd529nG | 25 | Detect Everything with Few Examples
Few-shot object detection aims at detecting novel categories given only a few example images. It is a basic skill for a robot to perform tasks in open environments. Recent methods focus on finetuning strategies, with complicated procedures that prohibit a wider application. In this p... | [
-0.03763836249709129,
-0.0371696874499321,
-0.019035493955016136,
0.033276062458753586,
0.0031838486902415752,
-0.017782684415578842,
-0.019558249041438103,
0.026372089982032776,
0.008607432246208191,
0.01053621806204319,
-0.04639901593327522,
-0.016917435452342033,
-0.00213608518242836,
0... | |
corl_2024_IcOrwlXzMi | IcOrwlXzMi | corl | 2,024 | VLM-Grounder: A VLM Agent for Zero-Shot 3D Visual Grounding | 3D visual grounding is crucial for robots, requiring integration of natural language and 3D scene understanding. Traditional methods depend on supervised learning with 3D point clouds are limited by scarce datasets. Recently zero-shot methods leveraging LLMs have been proposed to address the data issue. While effective... | Runsen Xu;Zhiwei Huang;Tai Wang;Yilun Chen;Jiangmiao Pang;Dahua Lin | The Chinese University of Hong Kong;Zhejiang University;Shanghai Artificial Intelligence Laboratory;Shanghai AI Laboratory ;The Chinese University of Hong Kong;Shanghai AI Laboratory | Poster | main | 3D Visual Grounding;VLM Agent;Zero-Shot | https://github.com/OpenRobotLab/VLM-Grounder | https://openreview.net/forum?id=IcOrwlXzMi | 9 | VLM-Grounder: A VLM Agent for Zero-Shot 3D Visual Grounding
3D visual grounding is crucial for robots, requiring integration of natural language and 3D scene understanding. Traditional methods depend on supervised learning with 3D point clouds are limited by scarce datasets. Recently zero-shot methods leveraging LLMs h... | [
-0.08284973353147507,
-0.023691989481449127,
-0.015620100311934948,
0.016640372574329376,
-0.017814137041568756,
0.0271049365401268,
0.005751447286456823,
0.0028757236432284117,
0.00455059576779604,
-0.012812093831598759,
-0.03925791755318642,
-0.0043068137019872665,
0.03315433859825134,
0... | |
corl_2024_InT87E5sr4 | InT87E5sr4 | corl | 2,024 | Dreamitate: Real-World Visuomotor Policy Learning via Video Generation | A key challenge in manipulation is learning a policy that can robustly generalize to diverse visual environments. A promising mechanism for learning robust policies is to leverage video generative models, which are pretrained on large-scale datasets of internet videos. In this paper, we propose a visuomotor policy lear... | Junbang Liang;Ruoshi Liu;Ege Ozguroglu;Sruthi Sudhakar;Achal Dave;Pavel Tokmakov;Shuran Song;Carl Vondrick | Columbia University;Columbia University;Columbia University;Columbia University;Toyota Research Institute;Toyota Research Institute;Stanford University;Columbia University | Poster | main | Imitation Learning;Visuomotor Policy;Video Generation | https://dreamitate.cs.columbia.edu/ | https://openreview.net/forum?id=InT87E5sr4 | 26 | Dreamitate: Real-World Visuomotor Policy Learning via Video Generation
A key challenge in manipulation is learning a policy that can robustly generalize to diverse visual environments. A promising mechanism for learning robust policies is to leverage video generative models, which are pretrained on large-scale datasets... | [
-0.06054406613111496,
-0.024151258170604706,
0.0005628764047287405,
0.01089571975171566,
-0.04609018564224243,
-0.01533881388604641,
-0.008485203608870506,
0.04026438668370247,
0.0023390769492834806,
0.021219922229647636,
-0.011513328179717064,
-0.020464042201638222,
0.021367410197854042,
... | |
corl_2024_IsZb0wT3Kw | IsZb0wT3Kw | corl | 2,024 | ANAVI: Audio Noise Awareness using Visual of Indoor environments for NAVIgation | We propose Audio Noise Awareness using Visuals of Indoors for NAVIgation for quieter robot path planning. While humans are naturally aware of the noise they make and its impact on those around them, robots currently lack this awareness.
A key challenge in achieving audio awareness for robots is estimating how loud wi... | Vidhi Jain;Rishi Veerapaneni;Yonatan Bisk | Google;;Meta | Poster | main | Robots;Acoustic Noise;Vision;Learning | https://github.com/vidhiJain/anavi | https://openreview.net/forum?id=IsZb0wT3Kw | 0 | ANAVI: Audio Noise Awareness using Visual of Indoor environments for NAVIgation
We propose Audio Noise Awareness using Visuals of Indoors for NAVIgation for quieter robot path planning. While humans are naturally aware of the noise they make and its impact on those around them, robots currently lack this awareness.
A... | [
-0.0473044328391552,
-0.02437174692749977,
-0.011816883459687233,
0.047710321843624115,
-0.02994348481297493,
0.013034548610448837,
0.005271935369819403,
0.009155548177659512,
0.01702885888516903,
0.036400798708200455,
-0.021401382982730865,
-0.021050842478871346,
-0.0020755650475621223,
-... | |
corl_2024_Isp19rFFV4 | Isp19rFFV4 | corl | 2,024 | Multi-Strategy Deployment-Time Learning and Adaptation for Navigation under Uncertainty | We present an approach for performant point-goal navigation in unfamiliar partially-mapped environments. When deployed, our robot runs multiple strategies for deployment-time learning and visual domain adaptation in parallel and quickly selects the best-performing among them. Choosing between policies as they are learn... | Abhishek Paudel;Xuesu Xiao;Gregory J. Stein | George Mason University;George Mason University;George Mason University | Poster | main | policy selection;domain adaptation;navigation under uncertainty | https://openreview.net/forum?id=Isp19rFFV4 | 2 | Multi-Strategy Deployment-Time Learning and Adaptation for Navigation under Uncertainty
We present an approach for performant point-goal navigation in unfamiliar partially-mapped environments. When deployed, our robot runs multiple strategies for deployment-time learning and visual domain adaptation in parallel and qui... | [
-0.03185751661658287,
-0.02208687923848629,
-0.03767871856689453,
0.03674283251166344,
-0.03496465086936951,
-0.007262465078383684,
0.03704231604933739,
-0.02534375712275505,
0.029237037524580956,
0.041553281247615814,
-0.032886989414691925,
-0.028151411563158035,
-0.007136120926588774,
0.... | ||
corl_2024_IssXUYvVTg | IssXUYvVTg | corl | 2,024 | MaIL: Improving Imitation Learning with Selective State Space Models | This work introduces Mamba Imitation Learning (MaIL), a novel imitation learning (IL) architecture that offers a computationally efficient alternative to state-of-the-art (SoTA) Transformer policies. Transformer-based policies have achieved remarkable results due to their ability in handling human-recorded data with in... | Xiaogang Jia;Qian Wang;Atalay Donat;Bowen Xing;Ge Li;Hongyi Zhou;Onur Celik;Denis Blessing;Rudolf Lioutikov;Gerhard Neumann | Karlsruher Institut für Technologie;Karlsruher Institut für Technologie;Karlsruher Institut für Technologie;Karlsruher Institut für Technologie;Karlsruhe Institute of Technology;Karlsruher Institut für Technologie;Karlsruhe Institute of Technology;Karlsruher Institut für Technologie;Karlsruher Institut für Technologie;... | Poster | main | Imitation Learning;Sequence Models;Denoising Diffusion Policies | https://github.com/ALRhub/MaIL | https://openreview.net/forum?id=IssXUYvVTg | 7 | MaIL: Improving Imitation Learning with Selective State Space Models
This work introduces Mamba Imitation Learning (MaIL), a novel imitation learning (IL) architecture that offers a computationally efficient alternative to state-of-the-art (SoTA) Transformer policies. Transformer-based policies have achieved remarkable... | [
-0.06365672498941422,
-0.01910075545310974,
-0.009905479848384857,
0.009793342091143131,
-0.02992202527821064,
0.009615791030228138,
0.01953061670064926,
0.06085328757762909,
0.009400861337780952,
0.027062518522143364,
-0.039659298956394196,
-0.046424929052591324,
-0.006924491375684738,
0.... | |
corl_2024_JScswMfEQ0 | JScswMfEQ0 | corl | 2,024 | Mobility VLA: Multimodal Instruction Navigation with Long-Context VLMs and Topological Graphs | An elusive goal in navigation research is to build an intelligent agent that can understand multimodal instructions including natural language and image, and perform useful navigation. To achieve this, we study a widely useful category of navigation tasks we call Multimodal Instruction Navigation with demonstration Tou... | Zhuo Xu;Hao-Tien Lewis Chiang;Zipeng Fu;Mithun George Jacob;Tingnan Zhang;Tsang-Wei Edward Lee;Wenhao Yu;Connor Schenck;David Rendleman;Dhruv Shah;Fei Xia;Jasmine Hsu;Jonathan Hoech;Pete Florence;Sean Kirmani;Sumeet Singh;Vikas Sindhwani;Carolina Parada;Chelsea Finn;Peng Xu;Sergey Levine;Jie Tan | Google DeepMind;Google Deepmind;Stanford University;;Google;;Google;;;UC Berkeley;Google;New York University;;Google;Google DeepMind;Google Brain Robotics;Google;Google DeepMind;Google;Google;Google;Google | Poster | main | vision-language navigation;multimodal foundation models;long-context reasoning | https://openreview.net/forum?id=JScswMfEQ0 | 20 | Mobility VLA: Multimodal Instruction Navigation with Long-Context VLMs and Topological Graphs
An elusive goal in navigation research is to build an intelligent agent that can understand multimodal instructions including natural language and image, and perform useful navigation. To achieve this, we study a widely useful... | [
-0.08175727725028992,
-0.042032770812511444,
0.011801938526332378,
0.024292632937431335,
-0.0395011231303215,
0.00802308414131403,
-0.046686530113220215,
-0.026563668623566628,
0.0021430570632219315,
0.011141104623675346,
-0.0031319812405854464,
-0.004267498850822449,
0.03534996882081032,
... | ||
corl_2024_JZzaRY8m8r | JZzaRY8m8r | corl | 2,024 | KOI: Accelerating Online Imitation Learning via Hybrid Key-state Guidance | Online Imitation Learning methods struggle with the gap between extensive online exploration space and limited expert trajectories, which hinder efficient exploration due to inaccurate task-aware reward estimation.
Inspired by the findings from cognitive neuroscience that task decomposition could facilitate cogniti... | Jingxian Lu;Wenke Xia;Dong Wang;Zhigang Wang;Bin Zhao;Di Hu;Xuelong Li | Renmin University of China;Renmin University of China;Shanghai AI Laboratory;Northwest Polytechnical University Xi'an;Renmin University of China;Northwestern Polytechnical University;Shanghai AI Lab | Poster | main | Online Imitation Learning; Robotic Manipulation | https://github.com/GeWu-Lab/Keystate_Online_Imitation | https://openreview.net/forum?id=JZzaRY8m8r | 0 | KOI: Accelerating Online Imitation Learning via Hybrid Key-state Guidance
Online Imitation Learning methods struggle with the gap between extensive online exploration space and limited expert trajectories, which hinder efficient exploration due to inaccurate task-aware reward estimation.
Inspired by the findings fr... | [
-0.055256687104701996,
-0.040260013192892075,
-0.03521834686398506,
-0.012567504309117794,
-0.004347293172031641,
0.03397167846560478,
0.005357918329536915,
0.01210917066782713,
-0.013328338041901588,
0.020753340795636177,
-0.016582505777478218,
-0.05720001831650734,
-0.011595836840569973,
... | |
corl_2024_KAzku0Uyh1 | KAzku0Uyh1 | corl | 2,024 | Object-Centric Dexterous Manipulation from Human Motion Data | Manipulating objects to achieve desired goal states is a basic but important skill for dexterous manipulation. Human hand motions demonstrate proficient manipulation capability, providing valuable data for training robots with multi-finger hands. Despite this potential, substantial challenges arise due to the embodimen... | Yuanpei Chen;Chen Wang;Yaodong Yang;Karen Liu | PsiRobot;Computer Science Department, Stanford University;Peking University;Computer Science Department, Stanford University | Poster | main | Dexterous Manipulation;Reinforcement Learning;Learning from Human | https://openreview.net/forum?id=KAzku0Uyh1 | 17 | Object-Centric Dexterous Manipulation from Human Motion Data
Manipulating objects to achieve desired goal states is a basic but important skill for dexterous manipulation. Human hand motions demonstrate proficient manipulation capability, providing valuable data for training robots with multi-finger hands. Despite this... | [
-0.034400295466184616,
-0.04207150638103485,
-0.05489999055862427,
0.02118363231420517,
-0.025009995326399803,
-0.04650786891579628,
0.01036075595766306,
-0.001600094372406602,
-0.006834771484136581,
0.010194391943514347,
-0.02805999480187893,
-0.025342723354697227,
0.013336815871298313,
-... | ||
corl_2024_KPcX4jetMw | KPcX4jetMw | corl | 2,024 | Reasoning Grasping via Multimodal Large Language Model | Despite significant progress in robotic systems for operation within human-centric environments, existing models still heavily rely on explicit human commands to identify and manipulate specific objects. This limits their effectiveness in environments where understanding and acting on implicit human intentions are cruc... | Shiyu Jin;JINXUAN XU;Yutian Lei;Liangjun Zhang | ;Rutgers University, New Brunswick;;Research, Baidu | Poster | main | Robotics Grasping;Multimodal Large Language Model | https://openreview.net/forum?id=KPcX4jetMw | 25 | Reasoning Grasping via Multimodal Large Language Model
Despite significant progress in robotic systems for operation within human-centric environments, existing models still heavily rely on explicit human commands to identify and manipulate specific objects. This limits their effectiveness in environments where underst... | [
-0.05841958895325661,
-0.020245151594281197,
0.007012965157628059,
-0.012176972813904285,
-0.049903176724910736,
0.003361741080880165,
-0.011439257301390171,
-0.022953221574425697,
-0.018872439861297607,
-0.019180599600076675,
-0.020861471071839333,
-0.014016592875123024,
0.02069338411092758... | ||
corl_2024_KULBk5q24a | KULBk5q24a | corl | 2,024 | CoViS-Net: A Cooperative Visual Spatial Foundation Model for Multi-Robot Applications | Autonomous robot operation in unstructured environments is often underpinned by spatial understanding through vision. Systems composed of multiple concurrently operating robots additionally require access to frequent, accurate and reliable pose estimates. Classical vision-based methods to regress relative pose are comm... | Jan Blumenkamp;Steven Morad;Jennifer Gielis;Amanda Prorok | University of Cambridge;University of Cambridge;; | Poster | main | Multi-Robot Systems;Robot Perception;Foundation Models | https://github.com/proroklab/CoViS-Net | https://openreview.net/forum?id=KULBk5q24a | 4 | CoViS-Net: A Cooperative Visual Spatial Foundation Model for Multi-Robot Applications
Autonomous robot operation in unstructured environments is often underpinned by spatial understanding through vision. Systems composed of multiple concurrently operating robots additionally require access to frequent, accurate and rel... | [
-0.05140884220600128,
-0.022277165204286575,
-0.03504645824432373,
0.04042688012123108,
-0.0009253360913135111,
0.020195014774799347,
0.020342422649264336,
0.00415739044547081,
0.04097966477274895,
0.017983881756663322,
-0.03976354002952576,
-0.008637238293886185,
0.013718237169086933,
0.0... | |
corl_2024_KXsropnmNI | KXsropnmNI | corl | 2,024 | Transferable Tactile Transformers for Representation Learning Across Diverse Sensors and Tasks | This paper presents T3: Transferable Tactile Transformers, a framework for tactile representation learning that scales across multi-sensors and multi-tasks.T3 is designed to overcome the contemporary issue that camera-based tactile sensing is extremely heterogeneous, i.e. sensors are built into different form factors, ... | Jialiang Zhao;Yuxiang Ma;Lirui Wang;Edward Adelson | Massachusetts Institute of Technology;Massachusetts Institute of Technology;Massachusetts Institute of Technology;Massachusetts Institute of Technology | Poster | main | Tactile Sensing;Representation Learning;Heterogeneous Learning;Robot Manipulation;Robot Learning | https://github.com/alanzjl/t3 | https://openreview.net/forum?id=KXsropnmNI | 18 | Transferable Tactile Transformers for Representation Learning Across Diverse Sensors and Tasks
This paper presents T3: Transferable Tactile Transformers, a framework for tactile representation learning that scales across multi-sensors and multi-tasks.T3 is designed to overcome the contemporary issue that camera-based t... | [
-0.023850975558161736,
-0.011649435386061668,
-0.04814363643527031,
0.05962743982672691,
0.02574653923511505,
0.0033563438337296247,
-0.008824492804706097,
0.008286189287900925,
0.0045755780301988125,
0.012965287081897259,
0.00726479385048151,
0.037800852209329605,
0.02447669580578804,
0.0... | |
corl_2024_KcW31O0PtL | KcW31O0PtL | corl | 2,024 | Hint-AD: Holistically Aligned Interpretability in End-to-End Autonomous Driving | End-to-end architectures in autonomous driving (AD) face a significant challenge in interpretability, impeding human-AI trust. Human-friendly natural language has been explored for tasks such as driving explanation and 3D captioning. However, previous works primarily focused on the paradigm of declarative interpretabil... | Kairui Ding;Boyuan Chen;Yuchen Su;Huan-ang Gao;Bu Jin;Chonghao Sima;Xiaohui Li;Wuqiang Zhang;Paul Barsch;Hongyang Li;Hao Zhao | Tsinghua University;Tsinghua University;Tsinghua University;Tsinghua University;;Shanghai AI Lab;Dalian University of Technology;Karlsruher Institut für Technologie;Technische Universität Dresden;Shanghai AI Lab;Peking University | Poster | main | Interpretability;Language alignment;Autonomous driving | https://openreview.net/forum?id=KcW31O0PtL | 5 | Hint-AD: Holistically Aligned Interpretability in End-to-End Autonomous Driving
End-to-end architectures in autonomous driving (AD) face a significant challenge in interpretability, impeding human-AI trust. Human-friendly natural language has been explored for tasks such as driving explanation and 3D captioning. Howeve... | [
-0.03900745138525963,
-0.03401169553399086,
-0.007249375805258751,
-0.010056031867861748,
-0.046971164643764496,
-0.023190628737211227,
0.011134451255202293,
-0.004484197590500116,
-0.009816383011639118,
0.02359618991613388,
-0.033956389874219894,
-0.05331264063715935,
0.020075194537639618,
... | ||
corl_2024_KdVLK0Wo5z | KdVLK0Wo5z | corl | 2,024 | PoliFormer: Scaling On-Policy RL with Transformers Results in Masterful Navigators | We present PoliFormer (Policy Transformer), an RGB-only indoor navigation agent trained end-to-end with reinforcement learning at scale that generalizes to the real-world without adaptation despite being trained purely in simulation. PoliFormer uses a foundational vision transformer encoder with a causal transformer de... | Kuo-Hao Zeng;Zichen Zhang;Kiana Ehsani;Rose Hendrix;Jordi Salvador;Alvaro Herrasti;Ross Girshick;Aniruddha Kembhavi;Luca Weihs | Allen Institute for Artificial Intelligence;Allen Institute for Artificial Intelligence;Allen Institute for Artificial Intelligence;Allen Institute for Artificial Intelligence;Allen Institute for AI;Allen Institute for Artificial Intelligence;Allen Institute for Artificial Intelligence;Allen Institute for Artificial In... | Poster | main | Embodied Navigation;On-Policy RL;Transformer Policy | https://openreview.net/forum?id=KdVLK0Wo5z | 16 | PoliFormer: Scaling On-Policy RL with Transformers Results in Masterful Navigators
We present PoliFormer (Policy Transformer), an RGB-only indoor navigation agent trained end-to-end with reinforcement learning at scale that generalizes to the real-world without adaptation despite being trained purely in simulation. Pol... | [
-0.06781663745641708,
-0.0017474069027230144,
-0.03663143143057823,
0.020050298422574997,
-0.021318597719073296,
0.0018220125930383801,
-0.014827894978225231,
-0.013867346569895744,
0.008052759803831577,
0.016534501686692238,
-0.025571124628186226,
-0.02049793303012848,
0.026615606620907784,... | ||
corl_2024_Ke5xrnBFAR | Ke5xrnBFAR | corl | 2,024 | Gameplay Filters: Robust Zero-Shot Safety through Adversarial Imagination | Despite the impressive recent advances in learning-based robot control, ensuring robustness to out-of-distribution conditions remains an open challenge. Safety filters can, in principle, keep arbitrary control policies from incurring catastrophic failures by overriding unsafe actions, but existing solutions for complex... | Duy Phuong Nguyen;Kai-Chieh Hsu;Wenhao Yu;Jie Tan;Jaime Fernández Fisac | Princeton University;Princeton University;Google;Google;Princeton University | Poster | main | Robust Safety;Adversarial Reinforcement Learning;Game Theory | https://openreview.net/forum?id=Ke5xrnBFAR | 7 | Gameplay Filters: Robust Zero-Shot Safety through Adversarial Imagination
Despite the impressive recent advances in learning-based robot control, ensuring robustness to out-of-distribution conditions remains an open challenge. Safety filters can, in principle, keep arbitrary control policies from incurring catastrophic... | [
-0.05416349321603775,
-0.04557614400982857,
-0.02579922042787075,
0.009344782680273056,
-0.005576200317591429,
-0.012713736854493618,
0.016170980408787727,
0.014953509904444218,
-0.006273225415498018,
0.02697022259235382,
-0.002261846326291561,
-0.01891261339187622,
0.0030994380358606577,
... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.