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_jPkOFAiOzf
jPkOFAiOzf
corl
2,024
Region-aware Grasp Framework with Normalized Grasp Space for Efficient 6-DoF Grasping
A series of region-based methods succeed in extracting regional features and enhancing grasp detection quality. However, faced with a cluttered scene with potential collision, the definition of the grasp-relevant region stays inconsistent. In this paper, we propose Normalized Grasp Space (NGS) from a novel region-aware...
Siang Chen;Pengwei Xie;Wei Tang;Dingchang Hu;Yixiang Dai;Guijin Wang
Tsinghua University;;;;;Department of Electronic Engineering, Tsinghua University
Poster
main
6-DoF Grasping;RGBD Perception;Normalized Space;Heatmap
https://github.com/THU-VCLab/RegionNormalizedGrasp
https://openreview.net/forum?id=jPkOFAiOzf
1
Region-aware Grasp Framework with Normalized Grasp Space for Efficient 6-DoF Grasping A series of region-based methods succeed in extracting regional features and enhancing grasp detection quality. However, faced with a cluttered scene with potential collision, the definition of the grasp-relevant region stays inconsis...
[ -0.05633707344532013, -0.01054462045431137, -0.02112640254199505, -0.0298593919724226, -0.01724300906062126, -0.033984337002038956, -0.018134890124201775, 0.009420480579137802, 0.014102849178016186, 0.02974790707230568, -0.026905041188001633, 0.020215941593050957, -0.01794908195734024, 0.0...
corl_2024_jart4nhCQr
jart4nhCQr
corl
2,024
Learning to Manipulate Anywhere: A Visual Generalizable Framework For Reinforcement Learning
Can we endow visuomotor robots with generalization capabilities to operate in diverse open-world scenarios? In this paper, we propose Maniwhere, a generalizable framework tailored for visual reinforcement learning, enabling the trained robot policies to generalize across a combination of multiple visual disturbance typ...
Zhecheng Yuan;Tianming Wei;Shuiqi Cheng;Gu Zhang;Yuanpei Chen;Huazhe Xu
;Shanghai Jiaotong University;University of Hong Kong;Shanghai Jiaotong University;PsiRobot;Tsinghua University
Poster
main
Visual Generalization;Sim2real;Reinforcement Learning
https://openreview.net/forum?id=jart4nhCQr
22
Learning to Manipulate Anywhere: A Visual Generalizable Framework For Reinforcement Learning Can we endow visuomotor robots with generalization capabilities to operate in diverse open-world scenarios? In this paper, we propose Maniwhere, a generalizable framework tailored for visual reinforcement learning, enabling the...
[ -0.05586991459131241, -0.01801093854010105, -0.0007039129268378019, -0.014964496716856956, -0.02187899872660637, -0.033104680478572845, -0.002679484896361828, 0.008474649861454964, 0.03631729260087013, 0.002393303904682398, -0.02132510021328926, -0.03277233988046646, -0.0009214335586875677, ...
corl_2024_jnubz7wB2w
jnubz7wB2w
corl
2,024
Verification of Neural Control Barrier Functions with Symbolic Derivative Bounds Propagation
Control barrier functions (CBFs) are important in safety-critical systems and robot control applications. Neural networks have been used to parameterize and synthesize CBFs with bounded control input for complex systems. However, it is still challenging to verify pre-trained neural networks CBFs (neural CBFs) in an eff...
Hanjiang Hu;Yujie Yang;Tianhao Wei;Changliu Liu
School of Computer Science, Carnegie Mellon University;Tsinghua University;Carnegie Mellon University;Carnegie Mellon University
Poster
main
Learning for control;control barrier function;formal verification
https://github.com/intelligent-control-lab/verify-neural-CBF
https://openreview.net/forum?id=jnubz7wB2w
8
Verification of Neural Control Barrier Functions with Symbolic Derivative Bounds Propagation Control barrier functions (CBFs) are important in safety-critical systems and robot control applications. Neural networks have been used to parameterize and synthesize CBFs with bounded control input for complex systems. Howeve...
[ -0.004828940145671368, -0.033423393964767456, -0.03965487331151962, -0.013184788636863232, 0.0009256991324946284, 0.00349035975523293, 0.02907414920628071, 0.009776662103831768, 0.021618301048874855, 0.02247718535363674, -0.03325892612338066, -0.021143173798918724, 0.015094435773789883, 0....
corl_2024_k0ogr4dnhG
k0ogr4dnhG
corl
2,024
ClutterGen: A Cluttered Scene Generator for Robot Learning
We introduce ClutterGen, a physically compliant simulation scene generator capable of producing highly diverse, cluttered, and stable scenes for robot learning. Generating such scenes is challenging as each object must adhere to physical laws like gravity and collision. As the number of objects increases, finding valid...
Yinsen Jia;Boyuan Chen
Duke University;Duke University
Poster
main
Simulation Scene Generation;Manipulation;Robot Learning
https://github.com/generalroboticslab/ClutterGen
https://openreview.net/forum?id=k0ogr4dnhG
4
ClutterGen: A Cluttered Scene Generator for Robot Learning We introduce ClutterGen, a physically compliant simulation scene generator capable of producing highly diverse, cluttered, and stable scenes for robot learning. Generating such scenes is challenging as each object must adhere to physical laws like gravity and c...
[ -0.03515755757689476, -0.003937774803489447, -0.02679983153939247, -0.019489116966724396, -0.026744725182652473, -0.04830949753522873, -0.040631409734487534, 0.030106185004115105, 0.028618324548006058, -0.010350722819566727, -0.009239421226084232, 0.010415012948215008, -0.015475570224225521,...
corl_2024_k4Nnxqcwt8
k4Nnxqcwt8
corl
2,024
Q-SLAM: Quadric Representations for Monocular SLAM
In this paper, we reimagine volumetric representations through the lens of quadrics. We posit that rigid scene components can be effectively decomposed into quadric surfaces. Leveraging this assumption, we reshape the volumetric representations with million of cubes by several quadric planes, which results in more accu...
Chensheng Peng;Chenfeng Xu;Yue Wang;Mingyu Ding;Heng Yang;Masayoshi Tomizuka;Kurt Keutzer;Marco Pavone;Wei Zhan
;University of California, Berkeley;NVIDIA;University of California, Berkeley;NVIDIA;;University of California, Berkeley;Stanford University;
Poster
main
Neural Radiance Fields;Simultaneous Localization and Mapping
https://github.com/PholyPeng/Q-SLAM
https://openreview.net/forum?id=k4Nnxqcwt8
6
Q-SLAM: Quadric Representations for Monocular SLAM In this paper, we reimagine volumetric representations through the lens of quadrics. We posit that rigid scene components can be effectively decomposed into quadric surfaces. Leveraging this assumption, we reshape the volumetric representations with million of cubes by...
[ -0.00484340637922287, 0.008612051606178284, -0.014032386243343353, -0.040682751685380936, 0.009170369245111942, -0.015046663582324982, 0.03202882781624794, -0.011091913096606731, 0.025161519646644592, 0.041352733969688416, 0.0013062309008091688, 0.0030451579950749874, -0.01928057335317135, ...
corl_2024_kEZXeaMrkD
kEZXeaMrkD
corl
2,024
Goal-Reaching Policy Learning from Non-Expert Observations via Effective Subgoal Guidance
In this work, we address the challenging problem of long-horizon goal-reaching policy learning from non-expert, action-free observation data. Unlike fully labeled expert data, our data is more accessible and avoids the costly process of action labeling. Additionally, compared to online learning, which often involves ai...
RenMing Huang;Shaochong Liu;Yunqiang Pei;Peng Wang;Guoqing Wang;Yang Yang;Heng Tao Shen
University of Electronic Science and Technology of China;University of Electronic Science and Technology of China;University of Electronic Science and Technology of China;University of Electronic Science and Technology of China;University of Electronic Science and Technology of China;University of Electronic Science an...
Poster
main
Goal-Reaching;Long-Horizon;Non-Expert Observation Data
https://github.com/RenMing-Huang/EGR-PO
https://openreview.net/forum?id=kEZXeaMrkD
1
Goal-Reaching Policy Learning from Non-Expert Observations via Effective Subgoal Guidance In this work, we address the challenging problem of long-horizon goal-reaching policy learning from non-expert, action-free observation data. Unlike fully labeled expert data, our data is more accessible and avoids the costly proc...
[ -0.07266918569803238, -0.04698599874973297, -0.023455560207366943, 0.012439122423529625, -0.03354538232088089, -0.04305490106344223, 0.0033461146522313356, -0.03137391805648804, 0.0019152411259710789, 0.027143307030200958, -0.01473226398229599, -0.02946452796459198, 0.0035075703635811806, ...
corl_2024_lKGRPJFPCM
lKGRPJFPCM
corl
2,024
InterACT: Inter-dependency Aware Action Chunking with Hierarchical Attention Transformers for Bimanual Manipulation
We present InterACT: Inter-dependency aware Action Chunking with Hierarchical Attention Transformers, a novel imitation learning framework for bimanual manipulation that integrates hierarchical attention to capture inter-dependencies between dual-arm joint states and visual inputs. InterACT consists of a Hierarchical A...
Andrew Choong-Won Lee;Ian Chuang;Ling-Yuan Chen;Iman Soltani
University of California, Davis;University of California, Davis;University of California, Davis;University of California, Davis
Poster
main
Robotics;Imitation Learning;Bimanual Manipulation
https://openreview.net/forum?id=lKGRPJFPCM
7
InterACT: Inter-dependency Aware Action Chunking with Hierarchical Attention Transformers for Bimanual Manipulation We present InterACT: Inter-dependency aware Action Chunking with Hierarchical Attention Transformers, a novel imitation learning framework for bimanual manipulation that integrates hierarchical attention ...
[ -0.041903305798769, -0.0371517688035965, -0.04108020290732384, 0.034570228308439255, -0.0025698512326925993, 0.0015655281022191048, 0.0026072647888213396, 0.0025862196926027536, 0.015900807455182076, 0.0031333945225924253, -0.04422295093536377, -0.02482396736741066, 0.0016555546317249537, ...
corl_2024_lpjPft4RQT
lpjPft4RQT
corl
2,024
TRANSIC: Sim-to-Real Policy Transfer by Learning from Online Correction
Learning in simulation and transferring the learned policy to the real world has the potential to enable generalist robots. The key challenge of this approach is to address simulation-to-reality (sim-to-real) gaps. Previous methods often require domain-specific knowledge *a priori*. We argue that a straightforward way ...
Yunfan Jiang;Chen Wang;Ruohan Zhang;Jiajun Wu;Li Fei-Fei
Stanford University;Computer Science Department, Stanford University;Stanford University;Stanford University;Stanford University
Poster
main
Sim-to-Real Transfer;Human-in-the-Loop;Robot Manipulation
https://github.com/transic-robot/transic
https://openreview.net/forum?id=lpjPft4RQT
28
TRANSIC: Sim-to-Real Policy Transfer by Learning from Online Correction Learning in simulation and transferring the learned policy to the real world has the potential to enable generalist robots. The key challenge of this approach is to address simulation-to-reality (sim-to-real) gaps. Previous methods often require do...
[ -0.08163098990917206, -0.023626107722520828, -0.0111433370038867, 0.029820984229445457, -0.020854225382208824, -0.04364318773150444, -0.0046740625984966755, 0.006413464900106192, -0.0031881292816251516, 0.048182375729084015, -0.01914272829890251, -0.0126967066898942, -0.0005034499918110669, ...
corl_2024_lt0Yf8Wh5O
lt0Yf8Wh5O
corl
2,024
Differentiable Robot Rendering
Vision foundation models trained on massive amounts of visual data have shown unprecedented reasoning and planning skills in open-world settings. A key challenge in applying them to robotic tasks is the modality gap between visual data and action data. We introduce differentiable robot rendering, a method allowing the ...
Ruoshi Liu;Alper Canberk;Shuran Song;Carl Vondrick
Columbia University;Columbia University;Stanford University;Columbia University
Poster
main
Robot Representation;Visual Foundation Model
https://github.com/cvlab-columbia/drrobot
https://openreview.net/forum?id=lt0Yf8Wh5O
6
Differentiable Robot Rendering Vision foundation models trained on massive amounts of visual data have shown unprecedented reasoning and planning skills in open-world settings. A key challenge in applying them to robotic tasks is the modality gap between visual data and action data. We introduce differentiable robot re...
[ -0.034049589186906815, -0.03291399031877518, 0.008081968873739243, 0.011502496898174286, -0.041028011590242386, -0.0059435670264065266, 0.00044616818195208907, 0.0015396951930597425, 0.03379316255450249, 0.012702199630439281, -0.03377484530210495, -0.0051697129383683205, 0.021155068650841713...
corl_2024_lyhS75loxe
lyhS75loxe
corl
2,024
A3VLM: Actionable Articulation-Aware Vision Language Model
Vision Language Models (VLMs) for robotics have received significant attention in recent years. As a VLM can understand robot observations and perform complex visual reasoning, it is regarded as a potential universal solution for general robotics challenges such as manipulation and navigation. However, previous robotic...
Siyuan Huang;Haonan Chang;Yuhan Liu;Yimeng Zhu;Hao Dong;Abdeslam Boularias;Peng Gao;Hongsheng Li
Shanghai Jiaotong University;Rutgers, New Brunswick;Rutgers University;Yuandao AI;Peking University;, Rutgers University;The Chinese University of Hong Kong;shanghai ai lab
Poster
main
LLM;VLM;Manipulation;Articulation
https://github.com/changhaonan/A3VLM
https://openreview.net/forum?id=lyhS75loxe
12
A3VLM: Actionable Articulation-Aware Vision Language Model Vision Language Models (VLMs) for robotics have received significant attention in recent years. As a VLM can understand robot observations and perform complex visual reasoning, it is regarded as a potential universal solution for general robotics challenges suc...
[ -0.02356579527258873, -0.02374918758869171, 0.016844499856233597, 0.009050365537405014, -0.016065087169408798, 0.005428385455161333, -0.009710574522614479, 0.004717743955552578, 0.020264750346541405, -0.01767893135547638, -0.019256098195910454, -0.014671312645077705, 0.020686550065875053, ...
corl_2024_ma7McOiCZY
ma7McOiCZY
corl
2,024
HYPERmotion: Learning Hybrid Behavior Planning for Autonomous Loco-manipulation
Enabling robots to autonomously perform hybrid motions in diverse environments can be beneficial for long-horizon tasks such as material handling, household chores, and work assistance. This requires extensive exploitation of intrinsic motion capabilities, extraction of affordances from rich environmental information, ...
Jin Wang;Rui Dai;Weijie Wang;Luca Rossini;Francesco Ruscelli;Nikos Tsagarakis
Istituto Italiano di Tecnologia;Università degli Studi di Genova, Istituto Italiano di Tecnologia;Università degli Studi di Genova, Istituto Italiano di Tecnologia;;;
Poster
main
Loco-manipulation;Large Language Models;Humanoid Robot Learning
https://openreview.net/forum?id=ma7McOiCZY
6
HYPERmotion: Learning Hybrid Behavior Planning for Autonomous Loco-manipulation Enabling robots to autonomously perform hybrid motions in diverse environments can be beneficial for long-horizon tasks such as material handling, household chores, and work assistance. This requires extensive exploitation of intrinsic moti...
[ -0.031588319689035416, -0.02839275822043419, 0.00655182171612978, 0.040256742388010025, -0.02126702107489109, 0.005587643478065729, -0.029751790687441826, 0.00045827170833945274, 0.0034182420931756496, -0.0013969108695164323, -0.03838348388671875, -0.02325047366321087, 0.0055049993097782135,...
corl_2024_nQslM6f7dW
nQslM6f7dW
corl
2,024
APRICOT: Active Preference Learning and Constraint-Aware Task Planning with LLMs
Home robots performing personalized tasks must adeptly balance user preferences with environmental affordances. We focus on organization tasks within constrained spaces, such as arranging items into a refrigerator, where preferences for placement collide with physical limitations. The robot must infer user preferences ...
Huaxiaoyue Wang;Nathaniel Chin;Gonzalo Gonzalez-Pumariega;Xiangwan Sun;Neha Sunkara;Maximus Adrian Pace;Jeannette Bohg;Sanjiban Choudhury
Cornell University;Cornell University;Cornell University;Cornell University;Cornell University;Cornell University;Stanford University;Cornell University
Poster
main
Active Preference Learning;Task Planning;Large Language Models
https://github.com/portal-cornell/apricot
https://openreview.net/forum?id=nQslM6f7dW
3
APRICOT: Active Preference Learning and Constraint-Aware Task Planning with LLMs Home robots performing personalized tasks must adeptly balance user preferences with environmental affordances. We focus on organization tasks within constrained spaces, such as arranging items into a refrigerator, where preferences for pl...
[ -0.01587836444377899, -0.04079081490635872, -0.023199087008833885, 0.005497414153069258, -0.05046626180410385, -0.035513296723365784, 0.001037064241245389, -0.057283055037260056, 0.02362055703997612, 0.027212200686335564, -0.028494929894804955, 0.010765769518911839, 0.0000856107726576738, ...
corl_2024_nVJm2RdPDu
nVJm2RdPDu
corl
2,024
DiffuseLoco: Real-Time Legged Locomotion Control with Diffusion from Offline Datasets
Offline learning at scale has led to breakthroughs in computer vision, natural language processing, and robotic manipulation domains. However, scaling up learning for legged robot locomotion, especially with multiple skills in a single policy, presents significant challenges for prior online reinforcement learning (RL)...
Xiaoyu Huang;Yufeng Chi;Ruofeng Wang;Zhongyu Li;Xue Bin Peng;Sophia Shao;Borivoje Nikolic;Koushil Sreenath
University of California, Berkeley;;University of California, Berkeley;University of California, Berkeley;Simon Fraser University;University of California, Berkeley;University of California, Berkeley;University of California, Berkeley
Poster
main
Offline Learning;Bipedal Walking;Imitation Learning
https://github.com/HybridRobotics/DiffuseLoco
https://openreview.net/forum?id=nVJm2RdPDu
29
DiffuseLoco: Real-Time Legged Locomotion Control with Diffusion from Offline Datasets Offline learning at scale has led to breakthroughs in computer vision, natural language processing, and robotic manipulation domains. However, scaling up learning for legged robot locomotion, especially with multiple skills in a singl...
[ -0.08259818702936172, -0.026999056339263916, -0.01109799463301897, 0.021268123760819435, -0.016565121710300446, 0.005071419756859541, -0.01685621589422226, 0.039297815412282944, 0.021013416349887848, -0.004894033540040255, -0.008660074323415756, -0.010661352425813675, -0.008801073767244816, ...
corl_2024_nmEt0ci8hi
nmEt0ci8hi
corl
2,024
General Flow as Foundation Affordance for Scalable Robot Learning
We address the challenge of acquiring real-world manipulation skills with a scalable framework. We hold the belief that identifying an appropriate prediction target capable of leveraging large-scale datasets is crucial for achieving efficient and universal learning. Therefore, we propose to utilize 3D flow, which repre...
Chengbo Yuan;Chuan Wen;Tong Zhang;Yang Gao
University of California, Berkeley;Tsinghua University;Tsinghua University;Wuhan University
Poster
main
Flow;Transferable Affordance;Scalability
https://github.com/michaelyuancb/general_flow
https://openreview.net/forum?id=nmEt0ci8hi
41
General Flow as Foundation Affordance for Scalable Robot Learning We address the challenge of acquiring real-world manipulation skills with a scalable framework. We hold the belief that identifying an appropriate prediction target capable of leveraging large-scale datasets is crucial for achieving efficient and univers...
[ -0.07066448777914047, -0.008810212835669518, -0.02681448496878147, -0.009294591844081879, 0.01595708727836609, 0.004930611699819565, -0.00754442997276783, -0.015646353363990784, 0.007114886771887541, 0.011277803219854832, -0.013544331304728985, -0.010738588869571686, 0.02776496484875679, 0...
corl_2024_oL1WEZQal8
oL1WEZQal8
corl
2,024
OmniH2O: Universal and Dexterous Human-to-Humanoid Whole-Body Teleoperation and Learning
We present OmniH2O (Omni Human-to-Humanoid), a learning-based system for whole-body humanoid teleoperation and autonomy. Using kinematic pose as a universal control interface, OmniH2O enables various ways for a human to control a full-sized humanoid with dexterous hands, including using real-time teleoperation through ...
Tairan He;Zhengyi Luo;Xialin He;Wenli Xiao;Chong Zhang;Weinan Zhang;Kris M. Kitani;Changliu Liu;Guanya Shi
Carnegie Mellon University;Meta Platforms, Inc.;Shanghai Jiaotong University;Carnegie Mellon University;ETHZ - ETH Zurich;Shanghai Jiaotong University;Carnegie Mellon University;Carnegie Mellon University;Carnegie Mellon University
Poster
main
Humanoid Teleoperation;Humanoid Loco-Manipulation;RL
https://github.com/LeCAR-Lab/human2humanoid
https://openreview.net/forum?id=oL1WEZQal8
69
OmniH2O: Universal and Dexterous Human-to-Humanoid Whole-Body Teleoperation and Learning We present OmniH2O (Omni Human-to-Humanoid), a learning-based system for whole-body humanoid teleoperation and autonomy. Using kinematic pose as a universal control interface, OmniH2O enables various ways for a human to control a f...
[ -0.021951092407107353, -0.0053414953872561455, -0.013410361483693123, 0.003873999696224928, -0.011890960857272148, -0.05707661435008049, -0.01887454092502594, 0.013438673689961433, 0.03318144381046295, 0.009564673528075218, -0.020573250949382782, -0.021214984357357025, -0.0009413677616976202...
corl_2024_oSU7M7MK6B
oSU7M7MK6B
corl
2,024
Learning Visuotactile Estimation and Control for Non-prehensile Manipulation under Occlusions
Manipulation without grasping, known as non-prehensile manipulation, is essential for dexterous robots in contact-rich environments, but presents many challenges relating with underactuation, hybrid-dynamics, and frictional uncertainty. Additionally, object occlusions in a scenario of contact uncertainty and where the ...
Juan Del Aguila Ferrandis;Joao Moura;Sethu Vijayakumar
;University of Edinburgh, University of Edinburgh;
Poster
main
State Estimation;Reinforcement Learning with Tactile Sensing;Non-prehensile Manipulation
https://openreview.net/forum?id=oSU7M7MK6B
4
Learning Visuotactile Estimation and Control for Non-prehensile Manipulation under Occlusions Manipulation without grasping, known as non-prehensile manipulation, is essential for dexterous robots in contact-rich environments, but presents many challenges relating with underactuation, hybrid-dynamics, and frictional un...
[ -0.04582994058728218, -0.005276844836771488, -0.020071782171726227, 0.032988518476486206, -0.0008314208243973553, -0.03319563716650009, 0.03720622882246971, 0.0008072960772551596, 0.05788053944706917, 0.030597228556871414, -0.02504265494644642, -0.03227301314473152, 0.01874433271586895, -0...
corl_2024_ovjxugn9Q2
ovjxugn9Q2
corl
2,024
SoftManiSim: A Fast Simulation Framework for Multi-Segment Continuum Manipulators Tailored for Robot Learning
This paper introduces SoftManiSim, a novel simulation framework for multi-segment continuum manipulators. Existing continuum robot simulators often rely on simplifying assumptions, such as constant curvature bending or ignoring contact forces, to meet real-time simulation and training demands. To bridge this gap, we pr...
Mohammadreza Kasaei;Hamidreza Kasaei;Mohsen Khadem
;University of Groningen;Edinburgh University, University of Edinburgh
Poster
main
Simulation Framework;Soft Robotics;Mathematical Modelling;Robot Learning
https://github.com/MohammadKasaei/SoftManiSim
https://openreview.net/forum?id=ovjxugn9Q2
1
SoftManiSim: A Fast Simulation Framework for Multi-Segment Continuum Manipulators Tailored for Robot Learning This paper introduces SoftManiSim, a novel simulation framework for multi-segment continuum manipulators. Existing continuum robot simulators often rely on simplifying assumptions, such as constant curvature be...
[ -0.04594612494111061, -0.0012468931963667274, -0.03372680023312569, -0.007730966899544001, -0.02200211212038994, -0.013767349533736706, 0.03108874522149563, 0.04151272773742676, 0.04484693333506584, 0.017183996737003326, -0.015159656293690205, -0.007749286480247974, -0.002306008245795965, ...
corl_2024_p6Wq6TjjHH
p6Wq6TjjHH
corl
2,024
Generative Factor Chaining: Coordinated Manipulation with Diffusion-based Factor Graph
Learning to plan for multi-step, multi-manipulator tasks is notoriously difficult because of the large search space and the complex constraint satisfaction problems. We present Generative Factor Chaining (GFC), a composable generative model for planning. GFC represents a planning problem as a spatial-temporal factor gr...
Utkarsh Aashu Mishra;Yongxin Chen;Danfei Xu
Toyota Research Institute;Georgia Institute of Technology;NVIDIA
Poster
main
Task and Motion Planning;Manipulation Planning;Bimanual Manipulation;Generative Models
https://openreview.net/forum?id=p6Wq6TjjHH
3
Generative Factor Chaining: Coordinated Manipulation with Diffusion-based Factor Graph Learning to plan for multi-step, multi-manipulator tasks is notoriously difficult because of the large search space and the complex constraint satisfaction problems. We present Generative Factor Chaining (GFC), a composable generativ...
[ -0.07111135870218277, -0.033857278525829315, -0.012018040753901005, 0.0009847732726484537, -0.0024206845555454493, -0.02473759651184082, -0.021137721836566925, 0.015147162601351738, 0.011371909640729427, 0.0454137958586216, -0.018839340656995773, -0.01940239779651165, -0.024442221969366074, ...
corl_2024_pPhTsonbXq
pPhTsonbXq
corl
2,024
GraspSplats: Efficient Manipulation with 3D Feature Splatting
The ability for robots to perform efficient and zero-shot grasping of object parts is crucial for practical applications and is becoming prevalent with recent advances in Vision-Language Models (VLMs). To bridge the 2D-to-3D gap for representations to support such a capability, existing methods rely on neural fields (N...
Mazeyu Ji;Ri-Zhao Qiu;Xueyan Zou;Xiaolong Wang
University of California, San Diego;University of California, San Diego;University of Wisconsin - Madison;University of California, San Diego
Poster
main
Zero-shot manipulation;Gaussian Splatting;Keypoint Tracking
https://github.com/jimazeyu/GraspSplats
https://openreview.net/forum?id=pPhTsonbXq
16
GraspSplats: Efficient Manipulation with 3D Feature Splatting The ability for robots to perform efficient and zero-shot grasping of object parts is crucial for practical applications and is becoming prevalent with recent advances in Vision-Language Models (VLMs). To bridge the 2D-to-3D gap for representations to suppor...
[ -0.04097266495227814, -0.030481457710266113, 0.011621170677244663, 0.008042952045798302, 0.0020291113760322332, -0.030407963320612907, 0.041634105145931244, -0.036379314959049225, 0.02044958434998989, 0.030940791592001915, -0.029360679909586906, 0.03737147897481918, 0.04611721262335777, 0....
corl_2024_pcPSGZFaCH
pcPSGZFaCH
corl
2,024
Modeling the Real World with High-Density Visual Particle Dynamics
We present High-Density Visual Particle Dynamics (HD-VPD), a learned world model that can emulate the physical dynamics of real scenes by processing massive latent point clouds containing 100K+ particles. To enable efficiency at this scale, we introduce a novel family of Point Cloud Transformers (PCTs) called Interlac...
William F Whitney;Jake Varley;Deepali Jain;Krzysztof Marcin Choromanski;Sumeet Singh;Vikas Sindhwani
Google DeepMind;Google;Google;Google Brain Robotics & Columbia University;Google Brain Robotics;Google
Poster
main
point clouds;particle dynamics;world models for control;Performers
https://openreview.net/forum?id=pcPSGZFaCH
1
Modeling the Real World with High-Density Visual Particle Dynamics We present High-Density Visual Particle Dynamics (HD-VPD), a learned world model that can emulate the physical dynamics of real scenes by processing massive latent point clouds containing 100K+ particles. To enable efficiency at this scale, we introduc...
[ -0.03918326646089554, -0.046835094690322876, -0.016421854496002197, 0.03896147385239601, -0.03295460715889931, 0.01212463341653347, 0.014712207950651646, 0.004158601630479097, 0.01940680667757988, 0.012420356273651123, -0.0294798631221056, -0.030718203634023666, 0.03727954998612404, 0.0280...
corl_2024_qUSa3F79am
qUSa3F79am
corl
2,024
Policy Adaptation via Language Optimization: Decomposing Tasks for Few-Shot Imitation
Learned language-conditioned robot policies often struggle to effectively adapt to new real-world tasks even when pre-trained across a diverse set of instructions. We propose a novel approach for few-shot adaptation to unseen tasks that exploits the semantic understanding of task decomposition provided by vision-langua...
Vivek Myers;Chunyuan Zheng;Oier Mees;Kuan Fang;Sergey Levine
University of California, Berkeley;University of California, Berkeley;Electrical Engineering & Computer Science Department, University of California, Berkeley;;Google
Poster
main
Reinforcement Learning;Vision-Language Models;Manipulation
https://github.com/vivekmyers/palo-robot
https://openreview.net/forum?id=qUSa3F79am
13
Policy Adaptation via Language Optimization: Decomposing Tasks for Few-Shot Imitation Learned language-conditioned robot policies often struggle to effectively adapt to new real-world tasks even when pre-trained across a diverse set of instructions. We propose a novel approach for few-shot adaptation to unseen tasks th...
[ -0.01834087260067463, -0.016021698713302612, 0.0061952415853738785, 0.051557715982198715, -0.040654830634593964, 0.0056916759349405766, 0.014469423331320286, 0.008676109835505486, -0.0000528397285961546, 0.037143729627132416, -0.032006438821554184, -0.015864623710513115, 0.02452225424349308,...
corl_2024_qoebyrnF36
qoebyrnF36
corl
2,024
Control with Patterns: A D-learning Method
Learning-based control policies are widely used in various tasks in the field of robotics and control. However, formal (Lyapunov) stability guarantees for learning-based controllers with nonlinear dynamical systems are challenging to obtain. We propose a novel control approach, namely Control with Patterns (CWP), to a...
Quan Quan;Kai-Yuan Cai;Chenyu Wang
Beihang University;Beihang University;Beihang University
Poster
main
Lyapunov Methods;Reinforcement Learning;Control with Patterns;D-learning;Visual Servoing
https://openreview.net/forum?id=qoebyrnF36
0
Control with Patterns: A D-learning Method Learning-based control policies are widely used in various tasks in the field of robotics and control. However, formal (Lyapunov) stability guarantees for learning-based controllers with nonlinear dynamical systems are challenging to obtain. We propose a novel control approac...
[ -0.049289483577013016, -0.05022725835442543, -0.03580426797270775, 0.006676961667835712, -0.047526463866233826, -0.028414597734808922, 0.02250661328434944, 0.03079654835164547, 0.021925192326307297, 0.017742713913321495, -0.014019744470715523, -0.01796777918934822, 0.03497902676463127, 0.0...
corl_2024_r6ZhiVYriY
r6ZhiVYriY
corl
2,024
Trust the PRoC3S: Solving Long-Horizon Robotics Problems with LLMs and Constraint Satisfaction
Recent developments in pretrained large language models (LLMs) applied to robotics have demonstrated their capacity for sequencing a set of discrete skills to achieve open-ended goals in simple robotic tasks. In this paper, we examine the topic of LLM planning for a set of *continuously parameterized* skills whose exec...
Aidan Curtis;Nishanth Kumar;Jing Cao;Tomás Lozano-Pérez;Leslie Pack Kaelbling
Massachusetts Institute of Technology;The AI Institute;Massachusetts Institute of Technology;Massachusetts Institute of Technology;Massachusetts Institute of Technology
Poster
main
LLMs for planning;task and motion planning;constraint satisfaction
https://github.com/Learning-and-Intelligent-Systems/proc3s
https://openreview.net/forum?id=r6ZhiVYriY
9
Trust the PRoC3S: Solving Long-Horizon Robotics Problems with LLMs and Constraint Satisfaction Recent developments in pretrained large language models (LLMs) applied to robotics have demonstrated their capacity for sequencing a set of discrete skills to achieve open-ended goals in simple robotic tasks. In this paper, w...
[ -0.0613127276301384, 0.0002893931814469397, -0.01652671955525875, -0.0038577953819185495, -0.015590362250804901, 0.01708853431046009, -0.019981881603598595, -0.037959955632686615, -0.017734622582793236, 0.013464828953146935, -0.01099284365773201, 0.014747640118002892, 0.004730949178338051, ...
corl_2024_rEteJcq61j
rEteJcq61j
corl
2,024
Toward General Object-level Mapping from Sparse Views with 3D Diffusion Priors
Object-level mapping builds a 3D map of objects in a scene with detailed shapes and poses from multi-view sensor observations. Conventional methods struggle to build complete shapes and estimate accurate poses due to partial occlusions and sensor noise. They require dense observations to cover all objects, which is c...
Ziwei Liao;Binbin Xu;Steven L. Waslander
University of Toronto;University of Toronto;University of Toronto
Poster
main
Mapping;Objects Reconstruction;Pose Estimation;Diffusion
https://github.com/TRAILab/GeneralObjectMapping
https://openreview.net/forum?id=rEteJcq61j
3
Toward General Object-level Mapping from Sparse Views with 3D Diffusion Priors Object-level mapping builds a 3D map of objects in a scene with detailed shapes and poses from multi-view sensor observations. Conventional methods struggle to build complete shapes and estimate accurate poses due to partial occlusions and...
[ -0.03474288433790207, -0.03290761634707451, -0.02921891026198864, -0.01062092836946249, 0.022750046104192734, -0.028492070734500885, -0.035615090280771255, -0.012565221637487411, -0.020296966657042503, 0.01050281710922718, -0.041139062494039536, 0.005033357068896294, 0.019515614956617355, ...
corl_2024_rRpmVq6yHv
rRpmVq6yHv
corl
2,024
SELFI: Autonomous Self-Improvement with RL for Vision-Based Navigation around People
Autonomous self-improving robots that interact and improve with experience are key to the real-world deployment of robotic systems. In this paper, we propose an online learning method, SELFI, that leverages online robot experience to rapidly fine-tune pre-trained control policies efficiently. SELFI applies online model...
Noriaki Hirose;Dhruv Shah;Kyle Stachowicz;Ajay Sridhar;Sergey Levine
Toyota Central R&D Labs., Inc;UC Berkeley;University of California, Berkeley;University of California, Berkeley;Google
Poster
main
online reinforcement learning;vision-based navigation
https://openreview.net/forum?id=rRpmVq6yHv
2
SELFI: Autonomous Self-Improvement with RL for Vision-Based Navigation around People Autonomous self-improving robots that interact and improve with experience are key to the real-world deployment of robotic systems. In this paper, we propose an online learning method, SELFI, that leverages online robot experience to r...
[ -0.06262903660535812, 0.0003240596561226994, 0.003905006218701601, -0.0015533919213339686, -0.024649478495121002, 0.0036839242093265057, 0.03730933368206024, 0.012054787017405033, 0.03442363068461418, 0.042969029396772385, -0.027553798630833626, -0.018757062032818794, 0.01971586048603058, ...
corl_2024_rThtgkXuvZ
rThtgkXuvZ
corl
2,024
NOD-TAMP: Generalizable Long-Horizon Planning with Neural Object Descriptors
Solving complex manipulation tasks in household and factory settings remains challenging due to long-horizon reasoning, fine-grained interactions, and broad object and scene diversity. Learning skills from demonstrations can be an effective strategy, but such methods often have limited generalizability beyond training ...
Shuo Cheng;Caelan Reed Garrett;Ajay Mandlekar;Danfei Xu
Georgia Institute of Technology;NVIDIA;NVIDIA;NVIDIA
Poster
main
Robot Learning;Robot Planning;Manipulation
https://openreview.net/forum?id=rThtgkXuvZ
0
NOD-TAMP: Generalizable Long-Horizon Planning with Neural Object Descriptors Solving complex manipulation tasks in household and factory settings remains challenging due to long-horizon reasoning, fine-grained interactions, and broad object and scene diversity. Learning skills from demonstrations can be an effective st...
[ -0.0031546105165034533, -0.014123227447271347, -0.013651848770678043, 0.00794091634452343, -0.006708079949021339, 0.004709253553301096, -0.014920945279300213, 0.00631375377997756, -0.00447809649631381, 0.017522592097520828, -0.0034832158125936985, -0.010705732740461826, -0.009708586148917675...
corl_2024_rY5T2aIjPZ
rY5T2aIjPZ
corl
2,024
DeliGrasp: Inferring Object Properties with LLMs for Adaptive Grasp Policies
Large language models (LLMs) can provide rich physical descriptions of most worldly objects, allowing robots to achieve more informed and capable grasping. We leverage LLMs' common sense physical reasoning and code-writing abilities to infer an object's physical characteristics-mass $m$, friction coefficient $\mu$, and...
William Xie;Maria Valentini;Jensen Lavering;Nikolaus Correll
;University of Colorado at Boulder;University of Colorado at Boulder;University of Colorado at Boulder
Poster
main
contact-rich manipulation;adaptive grasping;force control;produce manipulation
https://github.com/deligrasp/deligrasp
https://openreview.net/forum?id=rY5T2aIjPZ
4
DeliGrasp: Inferring Object Properties with LLMs for Adaptive Grasp Policies Large language models (LLMs) can provide rich physical descriptions of most worldly objects, allowing robots to achieve more informed and capable grasping. We leverage LLMs' common sense physical reasoning and code-writing abilities to infer a...
[ -0.03545873984694481, -0.004201515577733517, -0.004234157968312502, -0.006020151544362307, -0.062076110392808914, -0.05073528364300728, -0.026580065488815308, -0.02533033676445484, -0.01149937603622675, 0.012842368334531784, -0.02952718921005726, 0.00934965442866087, 0.003145307768136263, ...
corl_2024_rvKWXxIvj0
rvKWXxIvj0
corl
2,024
Non-rigid Relative Placement through 3D Dense Diffusion
The task of "relative placement" is to predict the placement of one object in relation to another, e.g. placing a mug on a mug rack. Recent methods for relative placement have made tremendous progress towards data-efficient learning for robot manipulation; using explicit object-centric geometric reasoning, these approa...
Eric Cai;Octavian Donca;Ben Eisner;David Held
Carnegie Mellon University;;Carnegie Mellon University;Carnegie Mellon University
Poster
main
Deformable;Non-rigid;Manipulation;Relative Placement
https://openreview.net/forum?id=rvKWXxIvj0
0
Non-rigid Relative Placement through 3D Dense Diffusion The task of "relative placement" is to predict the placement of one object in relation to another, e.g. placing a mug on a mug rack. Recent methods for relative placement have made tremendous progress towards data-efficient learning for robot manipulation; using e...
[ -0.025231435894966125, -0.016464892774820328, -0.014890230260789394, 0.012201332487165928, -0.02839917689561844, 0.023997489362955093, 0.011280477046966553, -0.009047403000295162, 0.03097757138311863, 0.0023642960004508495, -0.028822770342230797, -0.027128396555781364, 0.005584987346082926, ...
corl_2024_s0VNSnPeoA
s0VNSnPeoA
corl
2,024
Text2Interaction: Establishing Safe and Preferable Human-Robot Interaction
Adjusting robot behavior to human preferences can require intensive human feedback, preventing quick adaptation to new users and changing circumstances. Moreover, current approaches typically treat user preferences as a reward, which requires a manual balance between task success and user satisfaction. To integrate new...
Jakob Thumm;Christopher Agia;Marco Pavone;Matthias Althoff
Stanford University;Stanford University;Stanford University;Technische Universität München
Poster
main
Human-Robot Interaction;Human Preference Learning;Task and Motion Planning;Safe Control
https://github.com/JakobThumm/text2interaction
https://openreview.net/forum?id=s0VNSnPeoA
3
Text2Interaction: Establishing Safe and Preferable Human-Robot Interaction Adjusting robot behavior to human preferences can require intensive human feedback, preventing quick adaptation to new users and changing circumstances. Moreover, current approaches typically treat user preferences as a reward, which requires a ...
[ -0.030478840693831444, -0.022164734080433846, -0.03646276146173477, 0.04455316811800003, -0.004706976935267448, -0.02917393669486046, 0.020673414692282677, -0.016003722324967384, 0.02928578481078148, 0.014596289023756981, -0.02346963994204998, -0.021829187870025635, 0.0030059406999498606, ...
corl_2024_s0vHSq5QEv
s0vHSq5QEv
corl
2,024
Generalizing End-To-End Autonomous Driving In Real-World Environments Using Zero-Shot LLMs
Traditional autonomous driving methods adopt modular design, decomposing tasks into sub-tasks, including perception, prediction, planning, and control. In contrast, end-to-end autonomous driving directly outputs actions from raw sensor data, avoiding error accumulation. However, training an end-to-end model requires a ...
Zeyu Dong;Yimin Zhu;Yansong Li;Kevin Mahon;Yu Sun
State University of New York at Stony Brook;, State University of New York at Stony Brook;University of Illinois Chicago;Sunrise AI Tech;Sunrise Technology Inc.
Poster
main
End-to-end Autonomous Driving;Large Vision-Language Model;Generalization
https://openreview.net/forum?id=s0vHSq5QEv
5
Generalizing End-To-End Autonomous Driving In Real-World Environments Using Zero-Shot LLMs Traditional autonomous driving methods adopt modular design, decomposing tasks into sub-tasks, including perception, prediction, planning, and control. In contrast, end-to-end autonomous driving directly outputs actions from raw ...
[ -0.059785179793834686, -0.012708494439721107, -0.005797099322080612, -0.0021595230791717768, -0.044461313635110855, -0.022249074652791023, -0.021401841193437576, 0.006980462931096554, -0.01783793792128563, -0.00824670772999525, -0.03296841308474541, 0.001962679671123624, -0.00481633516028523...
corl_2024_s31IWg2kN5
s31IWg2kN5
corl
2,024
Exploring Under Constraints with Model-Based Actor-Critic and Safety Filters
Applying reinforcement learning (RL) to learn effective policies on physical robots without supervision remains challenging when it comes to tasks where safe exploration is critical. Constrained model-based RL (CMBRL) presents a promising approach to this problem. These methods are designed to learn constraint-adhering...
Ahmed Agha;Baris Kayalibay;Atanas Mirchev;Patrick van der Smagt;Justin Bayer
Volkswagen Group;Data Lab, Volkswagen Group;Machine Learning Research Lab, Volkswagen Group;Machine Learning Research Lab; Volkswagen Group;VW Group
Poster
main
Model-based RL;Safe RL;Safety Filter;Exploration
https://openreview.net/forum?id=s31IWg2kN5
3
Exploring Under Constraints with Model-Based Actor-Critic and Safety Filters Applying reinforcement learning (RL) to learn effective policies on physical robots without supervision remains challenging when it comes to tasks where safe exploration is critical. Constrained model-based RL (CMBRL) presents a promising appr...
[ -0.039608392864465714, -0.025339072570204735, -0.026313653215765953, -0.012687924318015575, -0.042660847306251526, -0.02486097626388073, 0.030432630330324173, 0.007874787785112858, 0.008651693351566792, 0.024474821984767914, -0.027508892118930817, -0.04438934847712517, 0.01916060596704483, ...
corl_2024_t0LkF9JnVb
t0LkF9JnVb
corl
2,024
PianoMime: Learning a Generalist, Dexterous Piano Player from Internet Demonstrations
In this work, we introduce PianoMime, a framework for training a piano-playing agent using internet demonstrations. The internet is a promising source of large-scale demonstrations for training our robot agents. In particular, for the case of piano-playing, Youtube is full of videos of professional pianists playing a ...
Cheng Qian;Julen Urain;Kevin Zakka;Jan Peters
Technische Universität München;University of California, Berkeley;TU Darmstadt;Technische Universität Darmstadt
Poster
main
Reinforcement Learning;Imitation Learning;Robotics;Dexterous Manipulation
https://github.com/sNiper-Qian/pianomime
https://openreview.net/forum?id=t0LkF9JnVb
7
PianoMime: Learning a Generalist, Dexterous Piano Player from Internet Demonstrations In this work, we introduce PianoMime, a framework for training a piano-playing agent using internet demonstrations. The internet is a promising source of large-scale demonstrations for training our robot agents. In particular, for th...
[ -0.031190406531095505, -0.005284783896058798, -0.04320380464196205, -0.0020708765368908644, -0.02241636998951435, -0.02149083837866783, 0.01536382082849741, 0.04490678012371063, -0.008315899409353733, 0.05923400819301605, 0.0003681879607029259, -0.015104671940207481, -0.023934241384267807, ...
corl_2024_tqsQGrmVEu
tqsQGrmVEu
corl
2,024
View-Invariant Policy Learning via Zero-Shot Novel View Synthesis
Large-scale visuomotor policy learning is a promising approach toward developing generalizable manipulation systems. Yet, policies that can be deployed on diverse embodiments, environments, and observational modalities remain elusive. In this work, we investigate how knowledge from large-scale visual data of the world...
Stephen Tian;Blake Wulfe;Kyle Sargent;Katherine Liu;Sergey Zakharov;Vitor Campagnolo Guizilini;Jiajun Wu
Stanford University;;Computer Science Department, Stanford University;Toyota Research Institute;Toyota Research Institute;Toyota Research Institute;Stanford University
Poster
main
generalization;visual imitation learning;view synthesis
https://github.com/s-tian/VISTA
https://openreview.net/forum?id=tqsQGrmVEu
10
View-Invariant Policy Learning via Zero-Shot Novel View Synthesis Large-scale visuomotor policy learning is a promising approach toward developing generalizable manipulation systems. Yet, policies that can be deployed on diverse embodiments, environments, and observational modalities remain elusive. In this work, we i...
[ -0.04042038694024086, 0.0013838697923347354, -0.010469573549926281, 0.047272536903619766, -0.00407301913946867, -0.005708605516701937, -0.0055400351993739605, 0.06443935632705688, 0.024456338956952095, 0.006565123796463013, -0.044575415551662445, -0.02729925327003002, 0.03311264514923096, ...
corl_2024_ty1cqzTtUv
ty1cqzTtUv
corl
2,024
RT-Sketch: Goal-Conditioned Imitation Learning from Hand-Drawn Sketches
Natural language and images are commonly used as goal representations in goal-conditioned imitation learning. However, language can be ambiguous and images can be over-specified. In this work, we study hand-drawn sketches as a modality for goal specification. Sketches can be easy to provide on the fly like language, bu...
Priya Sundaresan;Quan Vuong;Jiayuan Gu;Peng Xu;Ted Xiao;Sean Kirmani;Tianhe Yu;Michael Stark;Ajinkya Jain;Karol Hausman;Dorsa Sadigh;Jeannette Bohg;Stefan Schaal
Stanford University;physical intelligence;University of California, San Diego;Google;;Google DeepMind;Google Brain;;Intrinsic Innovation LLC;;Stanford University;Stanford University;
Poster
main
Visual Imitation Learning;Goal-Conditioned Manipulation
https://openreview.net/forum?id=ty1cqzTtUv
11
RT-Sketch: Goal-Conditioned Imitation Learning from Hand-Drawn Sketches Natural language and images are commonly used as goal representations in goal-conditioned imitation learning. However, language can be ambiguous and images can be over-specified. In this work, we study hand-drawn sketches as a modality for goal spe...
[ -0.019248105585575104, -0.011080323718488216, -0.015114465728402138, -0.017846105620265007, -0.013260208070278168, -0.011776801198720932, 0.008904961869120598, -0.007769794203341007, -0.0027949551586061716, 0.01582903414964676, -0.00918083917349577, -0.013576788827776909, -0.0071773361414670...
corl_2024_uEbJXWobif
uEbJXWobif
corl
2,024
EXTRACT: Efficient Policy Learning by Extracting Transferable Robot Skills from Offline Data
Most reinforcement learning (RL) methods focus on learning optimal policies over low-level action spaces. While these methods can perform well in their training environments, they lack the flexibility to transfer to new tasks. Instead, RL agents that can act over useful, temporally extended skills rather than low-leve...
Jesse Zhang;Minho Heo;Zuxin Liu;Erdem Biyik;Joseph J Lim;Yao Liu;Rasool Fakoor
NVIDIA;Korea Advanced Institute of Science & Technology;Salesforce AI Research;University of Southern California;Korea Advanced Institute of Science & Technology;Amazon;Amazon Web Services
Poster
main
reinforcement learning;skill-based reinformement learning;skill learning;transfer learning;foundation models for robotics;robot learning
https://openreview.net/forum?id=uEbJXWobif
2
EXTRACT: Efficient Policy Learning by Extracting Transferable Robot Skills from Offline Data Most reinforcement learning (RL) methods focus on learning optimal policies over low-level action spaces. While these methods can perform well in their training environments, they lack the flexibility to transfer to new tasks. ...
[ -0.06007625535130501, -0.02106713317334652, -0.006117005366832018, 0.02419227361679077, -0.004290176089853048, -0.016085289418697357, 0.01119536068290472, 0.014770890586078167, 0.02009282261133194, 0.00531274126842618, -0.04202396422624588, -0.008447074331343174, -0.01972516067326069, 0.04...
corl_2024_uHdVI3QMr6
uHdVI3QMr6
corl
2,024
A Dual Approach to Imitation Learning from Observations with Offline Datasets
Demonstrations are an effective alternative to task specification for learning agents in settings where designing a reward function is difficult. However, demonstrating expert behavior in the action space of the agent becomes unwieldy when robots have complex, unintuitive morphologies. We consider the practical setting...
Harshit Sikchi;Caleb Chuck;Amy Zhang;Scott Niekum
University of Texas, Austin;University of Texas, Austin;University of Massachusetts at Amherst;Meta Facebook
Poster
main
Learning from Observations;Imitation Learning
https://github.com/hari-sikchi/DILO
https://openreview.net/forum?id=uHdVI3QMr6
3
A Dual Approach to Imitation Learning from Observations with Offline Datasets Demonstrations are an effective alternative to task specification for learning agents in settings where designing a reward function is difficult. However, demonstrating expert behavior in the action space of the agent becomes unwieldy when ro...
[ -0.0622648261487484, -0.057512067258358, 0.006152796559035778, 0.018762344494462013, -0.03802207112312317, -0.01605437695980072, 0.004236955661326647, 0.0034586451947689056, 0.007539017591625452, 0.0012572704581543803, -0.043880123645067215, -0.03087451122701168, -0.009855527430772781, 0.0...
corl_2024_uJBMZ6S02T
uJBMZ6S02T
corl
2,024
Real-to-Sim Grasp: Rethinking the Gap between Simulation and Real World in Grasp Detection
For 6-DoF grasp detection, simulated data is expandable to train more powerful model, but it faces the challenge of the large gap between simulation and real world. Previous works bridge this gap with a sim-to-real way. However, this way explicitly or implicitly forces the simulated data to adapt to the noisy real data...
Jia-Feng Cai;Zibo Chen;Xiao-Ming Wu;Jian-Jian Jiang;Yi-Lin Wei;Wei-Shi Zheng
SUN YAT-SEN UNIVERSITY;SUN YAT-SEN UNIVERSITY;Macquarie University;SUN YAT-SEN UNIVERSITY;SUN YAT-SEN UNIVERSITY;SUN YAT-SEN UNIVERSITY
Poster
main
Grasp pose detection;simulated datasets;sim-to-real
https://openreview.net/forum?id=uJBMZ6S02T
4
Real-to-Sim Grasp: Rethinking the Gap between Simulation and Real World in Grasp Detection For 6-DoF grasp detection, simulated data is expandable to train more powerful model, but it faces the challenge of the large gap between simulation and real world. Previous works bridge this gap with a sim-to-real way. However, ...
[ -0.044450923800468445, 0.00032338325399905443, -0.01330357976257801, -0.017228461802005768, -0.01856161653995514, -0.04553236439824104, -0.01888791285455227, 0.01605379395186901, -0.006083101499825716, 0.017964961007237434, -0.04109473153948784, 0.013210352510213852, -0.02880733087658882, ...
corl_2024_uMZ2jnZUDX
uMZ2jnZUDX
corl
2,024
Learning H-Infinity Locomotion Control
Stable locomotion in precipitous environments is an essential task for quadruped robots, requiring the ability to resist various external disturbances. Recent neural policies enhance robustness against disturbances by learning to resist external forces sampled from a fixed distribution in the simulated environment. How...
Junfeng Long;Wenye Yu;Quanyi Li;ZiRui Wang;Dahua Lin;Jiangmiao Pang
Shanghai AI Laboratory;Shanghai Jiaotong University;University of Edinburgh;The Chinese University of Hong Kong;Shanghai AI Laboratory ;Shanghai Artificial Intelligence Laboratory
Poster
main
Robot Learning;Quadrupedal Robot;Robust Locomotion
https://openreview.net/forum?id=uMZ2jnZUDX
9
Learning H-Infinity Locomotion Control Stable locomotion in precipitous environments is an essential task for quadruped robots, requiring the ability to resist various external disturbances. Recent neural policies enhance robustness against disturbances by learning to resist external forces sampled from a fixed distrib...
[ -0.06603187322616577, -0.011805005371570587, -0.002105146646499634, 0.0033582383766770363, -0.03735712915658951, 0.014518252573907375, -0.020944364368915558, 0.008687151595950127, 0.050152234733104706, 0.0030083723831921816, -0.057159073650836945, -0.025475963950157166, 0.03699536249041557, ...
corl_2024_ubq7Co6Cbv
ubq7Co6Cbv
corl
2,024
Gaussian Splatting to Real World Flight Navigation Transfer with Liquid Networks
Simulators are powerful tools for autonomous robot learning as they offer scalable data generation, flexible design, and optimization of trajectories. However, transferring behavior learned from simulation data into the real world proves to be difficult, usually mitigated with compute-heavy domain randomization method...
Alex Quach;Makram Chahine;Alexander Amini;Ramin Hasani;Daniela Rus
Liquid AI;Massachusetts Institute of Technology;Massachusetts Institute of Technology;Massachusetts Institute of Technology;Massachusetts Institute of Technology
Poster
main
End-to-end learning;Gaussian Splatting;Sim-to-real transfer
https://github.com/alexquach/multienv_sim
https://openreview.net/forum?id=ubq7Co6Cbv
7
Gaussian Splatting to Real World Flight Navigation Transfer with Liquid Networks Simulators are powerful tools for autonomous robot learning as they offer scalable data generation, flexible design, and optimization of trajectories. However, transferring behavior learned from simulation data into the real world proves ...
[ -0.07163572311401367, -0.02361311949789524, 0.0006505182827822864, -0.02381684072315693, -0.01427899207919836, -0.03716982901096344, -0.023483477532863617, -0.005167106166481972, 0.011973240412771702, 0.048781927675008774, -0.02620593272149563, 0.011621358804404736, 0.00022122829977888614, ...
corl_2024_ueBmGhLOXP
ueBmGhLOXP
corl
2,024
EquiBot: SIM(3)-Equivariant Diffusion Policy for Generalizable and Data Efficient Learning
Building effective imitation learning methods that enable robots to learn from limited data and still generalize across diverse real-world environments is a long-standing problem in robot learning. We propose EquiBot, a robust, data-efficient, and generalizable approach for robot manipulation task learning. Our approac...
Jingyun Yang;Ziang Cao;Congyue Deng;Rika Antonova;Shuran Song;Jeannette Bohg
Stanford University;;Stanford University;;Stanford University;Stanford University
Poster
main
Imitation Learning;Equivariance;Data Efficiency
https://github.com/yjy0625/equibot
https://openreview.net/forum?id=ueBmGhLOXP
38
EquiBot: SIM(3)-Equivariant Diffusion Policy for Generalizable and Data Efficient Learning Building effective imitation learning methods that enable robots to learn from limited data and still generalize across diverse real-world environments is a long-standing problem in robot learning. We propose EquiBot, a robust, d...
[ -0.07530080527067184, -0.037038952112197876, -0.0011927876621484756, 0.00020410565775819123, -0.03005361743271351, -0.047841209918260574, -0.022642120718955994, 0.02770046703517437, 0.019436649978160858, 0.043579597026109695, -0.03396318107843399, 0.018028464168310165, -0.011923245154321194,...
corl_2024_vBj5oC60Lk
vBj5oC60Lk
corl
2,024
Lifelong Autonomous Improvement of Navigation Foundation Models in the Wild
Recent works have proposed a number of general-purpose robotic foundation models that can control a variety of robotic platforms to perform a range of different tasks, including in the domains of navigation and manipulation. However, such models are typically trained via imitation learning, which precludes the ability ...
Kyle Stachowicz;Lydia Ignatova;Sergey Levine
University of California, Berkeley;University of California, Berkeley;Google
Poster
main
Navigation;Reinforcement Learning;Lifelong Learning
https://github.com/kylestach/lifelong-nav-rl
https://openreview.net/forum?id=vBj5oC60Lk
2
Lifelong Autonomous Improvement of Navigation Foundation Models in the Wild Recent works have proposed a number of general-purpose robotic foundation models that can control a variety of robotic platforms to perform a range of different tasks, including in the domains of navigation and manipulation. However, such model...
[ -0.058096107095479965, -0.011562192812561989, -0.004548531491309404, -0.01821252331137657, -0.00883951224386692, -0.012307249940931797, 0.02496403455734253, 0.010982703417539597, -0.0013084501260891557, 0.017531853169202805, -0.03405189886689186, -0.015204696916043758, -0.06361505389213562, ...
corl_2024_vhGkyWgctu
vhGkyWgctu
corl
2,024
Learning Decentralized Multi-Biped Control for Payload Transport
Payload transport over flat terrain via multi-wheel robot carriers is well-understood, highly effective, and configurable. In this paper, our goal is to provide similar effectiveness and configurability for transport over rough terrain that is more suitable for legs rather than wheels. For this purpose, we consider mul...
Bikram Pandit;Ashutosh Gupta;Mohitvishnu S. Gadde;Addison Johnson;Aayam Kumar Shrestha;Helei Duan;Jeremy Dao;Alan Fern
Oregon State University;Oregon State University;Oregon State University;;Oregon State University;;Oregon State University;Oregon State University
Poster
main
Multi-robot Transport;Bipedal locomotion;Reinforcement Learning
https://github.com/osudrl/roadrunner/tree/paper/decmbc
https://openreview.net/forum?id=vhGkyWgctu
4
Learning Decentralized Multi-Biped Control for Payload Transport Payload transport over flat terrain via multi-wheel robot carriers is well-understood, highly effective, and configurable. In this paper, our goal is to provide similar effectiveness and configurability for transport over rough terrain that is more suitab...
[ -0.021763188764452934, -0.03987745940685272, -0.018291130661964417, 0.02865145169198513, -0.03704768419265747, -0.019035808742046356, 0.030420059338212013, 0.01670869253575802, -0.0009349188185296953, 0.027720604091882706, -0.03203973174095154, 0.016057100147008896, 0.022023824974894524, 0...
corl_2024_vobaOY0qDl
vobaOY0qDl
corl
2,024
Jacta: A Versatile Planner for Learning Dexterous and Whole-body Manipulation
Robotic manipulation is challenging due to discontinuous dynamics, as well as high-dimensional state and action spaces. Data-driven approaches that succeed in manipulation tasks require large amounts of data and expert demonstrations, typically from humans. Existing planners are restricted to specific systems and often...
Jan Bruedigam;Ali Adeeb Abbas;Maks Sorokin;Kuan Fang;Brandon Hung;Maya Guru;Stefan Georg Sosnowski;Jiuguang Wang;Sandra Hirche;Simon Le Cleac'h
Technische Universität München;;;;;;;;Technical University Munich;
Poster
main
Dexterous Manipulation Planning;Learning with Demonstrations
https://openreview.net/forum?id=vobaOY0qDl
2
Jacta: A Versatile Planner for Learning Dexterous and Whole-body Manipulation Robotic manipulation is challenging due to discontinuous dynamics, as well as high-dimensional state and action spaces. Data-driven approaches that succeed in manipulation tasks require large amounts of data and expert demonstrations, typical...
[ -0.04443906992673874, -0.0074080717749893665, -0.03560366854071617, 0.03206576406955719, -0.026843145489692688, -0.022743668407201767, 0.02096535637974739, 0.016210714355111122, 0.025982066988945007, 0.011456070467829704, -0.006252169143408537, -0.008227030746638775, -0.02418503724038601, ...
corl_2024_vtEn8NJWlz
vtEn8NJWlz
corl
2,024
Learning Robotic Manipulation Policies from Point Clouds with Conditional Flow Matching
Learning from expert demonstrations is a popular approach to train robotic manipulation policies from limited data. However, imitation learning algorithms require a number of design choices ranging from the input modality, training objective, and 6-DoF end-effector pose representation. Diffusion-based methods have gain...
Eugenio Chisari;Nick Heppert;Max Argus;Tim Welschehold;Thomas Brox;Abhinav Valada
Universität Freiburg;University of Freiburg, Albert-Ludwigs-Universität Freiburg;Universität Freiburg;University of Freiburg;University of Freiburg;University of Freiburg, Albert-Ludwigs-Universität Freiburg
Poster
main
Imitation Learning;Manipulation;Conditional Flow Matching
https://openreview.net/forum?id=vtEn8NJWlz
14
Learning Robotic Manipulation Policies from Point Clouds with Conditional Flow Matching Learning from expert demonstrations is a popular approach to train robotic manipulation policies from limited data. However, imitation learning algorithms require a number of design choices ranging from the input modality, training ...
[ -0.03637746348977089, -0.023039670661091805, -0.028795020654797554, -0.02601783536374569, -0.0187094546854496, -0.02466578409075737, -0.0033892616629600525, 0.009564843028783798, 0.032248228788375854, 0.03117024339735508, -0.019093144685029984, -0.046554386615753174, 0.010350494645535946, ...
corl_2024_wD2kUVLT1g
wD2kUVLT1g
corl
2,024
Equivariant Diffusion Policy
Recent work has shown diffusion models are an effective approach to learning the multimodal distributions arising from demonstration data in behavior cloning. However, a drawback of this approach is the need to learn a denoising function, which is significantly more complex than learning an explicit policy. In this wor...
Dian Wang;Stephen Hart;David Surovik;Tarik Kelestemur;Haojie Huang;Haibo Zhao;Mark Yeatman;Jiuguang Wang;Robin Walters;Robert Platt
Northeastern University;The Robotics & AI Institute;The AI Institute ;Boston Dynamics AI Institute;Northeastern University;Northeastern University;;;Northeastern University ;Northeastern University
Poster
main
Equivariance;Diffusion Model;Robotic Manipulation
https://openreview.net/forum?id=wD2kUVLT1g
26
Equivariant Diffusion Policy Recent work has shown diffusion models are an effective approach to learning the multimodal distributions arising from demonstration data in behavior cloning. However, a drawback of this approach is the need to learn a denoising function, which is significantly more complex than learning an...
[ -0.07286377251148224, -0.017805056646466255, -0.03369264677166939, -0.001129365642555058, -0.03288913518190384, -0.02419661544263363, -0.0034423109609633684, 0.04262256622314453, 0.04196514934301376, 0.03977375850081444, -0.03190300986170769, -0.0012988561065867543, -0.03153777867555618, 0...
corl_2024_wH7Wv0nAm8
wH7Wv0nAm8
corl
2,024
Bi-Level Motion Imitation for Humanoid Robots
Imitation learning from human motion capture (MoCap) data provides a promising way to train humanoid robots. However, due to differences in morphology, such as varying degrees of joint freedom and force limits, exact replication of human behaviors may not be feasible for humanoid robots. Consequently, incorporating phy...
Wenshuai Zhao;Yi Zhao;Joni Pajarinen;Michael Muehlebach
Aalto University;Max Planck Institute for Intelligent Systems;Aalto University;Max-Planck Institute
Poster
main
Humanoid Robots;Imitation Learning;Latent Dynamics Model
https://openreview.net/forum?id=wH7Wv0nAm8
2
Bi-Level Motion Imitation for Humanoid Robots Imitation learning from human motion capture (MoCap) data provides a promising way to train humanoid robots. However, due to differences in morphology, such as varying degrees of joint freedom and force limits, exact replication of human behaviors may not be feasible for hu...
[ -0.067738838493824, -0.06910992413759232, -0.022715486586093903, 0.003890907159075141, -0.025216784328222275, 0.011607873253524303, 0.006684022955596447, 0.013960945419967175, 0.000352034461684525, 0.007124065887182951, -0.05943823978304863, -0.02584674023091793, -0.0032099983654916286, 0....
corl_2024_wSWMsjuMTI
wSWMsjuMTI
corl
2,024
ManiWAV: Learning Robot Manipulation from In-the-Wild Audio-Visual Data
Audio signals provide rich information for the robot interaction and object properties through contact. These information can surprisingly ease the learning of contact-rich robot manipulation skills, especially when the visual information alone is ambiguous or incomplete. However, the usage of audio data in robot manip...
Zeyi Liu;Cheng Chi;Eric Cousineau;Naveen Kuppuswamy;Benjamin Burchfiel;Shuran Song
Stanford University;Stanford University;Toyota Research Institute;Toyota Research Institute;Dexterous Manipulation Group, Toyota Research Institute;Stanford University
Poster
main
Robot Manipulation;Imitation Learning;Audio
https://github.com/real-stanford/maniwav
https://openreview.net/forum?id=wSWMsjuMTI
24
ManiWAV: Learning Robot Manipulation from In-the-Wild Audio-Visual Data Audio signals provide rich information for the robot interaction and object properties through contact. These information can surprisingly ease the learning of contact-rich robot manipulation skills, especially when the visual information alone is ...
[ -0.04663747176527977, -0.011349578388035297, -0.031448375433683395, -0.0025088293477892876, -0.026078684255480766, -0.01616540364921093, 0.014541354030370712, 0.03142959997057915, 0.003072083927690983, 0.04855253919959068, -0.005515200551599264, -0.04250694066286087, -0.008049845695495605, ...
corl_2024_wTKJge0PTq
wTKJge0PTq
corl
2,024
HiRT: Enhancing Robotic Control with Hierarchical Robot Transformers
Large Vision-Language-Action (VLA) models, leveraging powerful pre-trained Vision-Language Models (VLMs) backends, have shown promise in robotic control due to their impressive generalization ability. However, the success comes at a cost. Their reliance on VLM backends with billions of parameters leads to high computat...
Jianke Zhang;Yanjiang Guo;Xiaoyu Chen;Yen-Jen Wang;Yucheng Hu;Chengming Shi;Jianyu Chen
Beijing Institute of Technology;Tsinghua University;Tsinghua University;Tsinghua University;Tsinghua University;;Tsinghua University
Poster
main
Imitation Learning;Robots;Vision Language Models
https://openreview.net/forum?id=wTKJge0PTq
8
HiRT: Enhancing Robotic Control with Hierarchical Robot Transformers Large Vision-Language-Action (VLA) models, leveraging powerful pre-trained Vision-Language Models (VLMs) backends, have shown promise in robotic control due to their impressive generalization ability. However, the success comes at a cost. Their relian...
[ -0.04792465642094612, -0.013816536404192448, 0.00029638473642989993, 0.02772623859345913, -0.017431382089853287, 0.04088129475712776, 0.0065402621403336525, -0.0002752768050413579, 0.01597798988223076, 0.01155260019004345, -0.03760184720158577, 0.007020067889243364, 0.020235681906342506, 0...
corl_2024_wcbrhPnOei
wcbrhPnOei
corl
2,024
RobotKeyframing: Learning Locomotion with High-Level Objectives via Mixture of Dense and Sparse Rewards
This paper presents a novel learning-based control framework that uses keyframing to incorporate high-level objectives in natural locomotion for legged robots. These high-level objectives are specified as a variable number of partial or complete pose targets that are spaced arbitrarily in time. Our proposed framework u...
Fatemeh Zargarbashi;Jin Cheng;Dongho Kang;Robert Sumner;Stelian Coros
Disney Research|Studios;ETHZ - ETH Zurich;ETHZ - ETH Zurich;Disney Research, Disney Research;ETHZ - ETH Zurich
Poster
main
Legged robots;Multi-Critic Reinforcement Learning;Motion Imitation
https://openreview.net/forum?id=wcbrhPnOei
7
RobotKeyframing: Learning Locomotion with High-Level Objectives via Mixture of Dense and Sparse Rewards This paper presents a novel learning-based control framework that uses keyframing to incorporate high-level objectives in natural locomotion for legged robots. These high-level objectives are specified as a variable ...
[ -0.032561045140028, -0.0358959399163723, -0.010581881739199162, 0.005996399559080601, -0.026184430345892906, 0.005927686113864183, 0.017040951177477837, -0.015785785391926765, 0.0084975715726614, -0.00023663217143621296, -0.03217624872922897, -0.033055782318115234, 0.024425365030765533, 0....
corl_2024_xYJn2e1uu8
xYJn2e1uu8
corl
2,024
Sparsh: Self-supervised touch representations for vision-based tactile sensing
In this work, we introduce general purpose touch representations for the increasingly accessible class of vision-based tactile sensors. Such sensors have led to many recent advances in robot manipulation as they markedly complement vision, yet solutions today often rely on task and sensor specific handcrafted perceptio...
Carolina Higuera;Akash Sharma;Chaithanya Krishna Bodduluri;Taosha Fan;Patrick Lancaster;Mrinal Kalakrishnan;Michael Kaess;Byron Boots;Mike Lambeta;Tingfan Wu;Mustafa Mukadam
University of Washington;Carnegie Mellon University;Meta Facebook;;Meta;Meta;Carnegie Mellon University;;Meta;;Meta AI
Poster
main
Tactile sensing;Pre-trained representations;Self-supervised learning
https://github.com/facebookresearch/sparsh
https://openreview.net/forum?id=xYJn2e1uu8
10
Sparsh: Self-supervised touch representations for vision-based tactile sensing In this work, we introduce general purpose touch representations for the increasingly accessible class of vision-based tactile sensors. Such sensors have led to many recent advances in robot manipulation as they markedly complement vision, y...
[ -0.023849645629525185, 0.0018336874200031161, -0.02879568375647068, 0.002140472875908017, -0.014191756956279278, 0.008561891503632069, 0.022107291966676712, -0.025029949843883514, -0.0060560861602425575, 0.06519777327775955, 0.009920178912580013, 0.03945589438080788, 0.005943676456809044, ...
corl_2024_xYleTh2QhS
xYleTh2QhS
corl
2,024
Adaptive Diffusion Terrain Generator for Autonomous Uneven Terrain Navigation
Model-free reinforcement learning has emerged as a powerful method for developing robust robot control policies capable of navigating through complex and unstructured terrains. The effectiveness of these methods hinges on two essential elements: (1) the use of massively parallel physics simulations to expedite policy ...
Youwei Yu;Junhong Xu;Lantao Liu
Indiana University;Indiana University, Bloomington;Indiana University, Bloomington
Poster
main
Curriculum Reinforcement Learning;Diffusion Model;Field Robotics
https://openreview.net/forum?id=xYleTh2QhS
3
Adaptive Diffusion Terrain Generator for Autonomous Uneven Terrain Navigation Model-free reinforcement learning has emerged as a powerful method for developing robust robot control policies capable of navigating through complex and unstructured terrains. The effectiveness of these methods hinges on two essential elemen...
[ -0.043289899826049805, -0.03520277515053749, -0.00006625392416026443, 0.010063163936138153, -0.02221214771270752, -0.06564842164516449, -0.004038988146930933, 0.0203641839325428, 0.030262678861618042, 0.010081460699439049, -0.042521439492702484, -0.008654320612549782, -0.03249486908316612, ...
corl_2024_xcBH8Jhmbi
xcBH8Jhmbi
corl
2,024
Discovering Robotic Interaction Modes with Discrete Representation Learning
Abstract: Human actions manipulating articulated objects, such as opening and closing a drawer, can be categorized into multiple modalities we define as interaction modes. Traditional robot learning approaches lack discrete representations of these modes, which are crucial for empirical sampling and grounding. In this ...
Liquan Wang;Ankit Goyal;Haoping Xu;Animesh Garg
Department of Computer Science;NVIDIA;Toronto University;NVIDIA
Poster
main
Discovering Robotic Interaction Modes with Discrete Representation Learning
https://github.com/pairlab/ActAIM.git
https://openreview.net/forum?id=xcBH8Jhmbi
1
Discovering Robotic Interaction Modes with Discrete Representation Learning Abstract: Human actions manipulating articulated objects, such as opening and closing a drawer, can be categorized into multiple modalities we define as interaction modes. Traditional robot learning approaches lack discrete representations of t...
[ -0.011820705607533455, -0.054956406354904175, 0.00026903976686298847, 0.03451232239603996, -0.030299372971057892, 0.0038062294479459524, 0.01316546369343996, -0.006347633898258209, 0.0031526582315564156, 0.03665641322731972, -0.01765112392604351, -0.029434215277433395, -0.010071579366922379,...
corl_2024_xeFKtSXPMd
xeFKtSXPMd
corl
2,024
OCCAM: Online Continuous Controller Adaptation with Meta-Learned Models
Control tuning and adaptation present a significant challenge to the usage of robots in diverse environments. It is often nontrivial to find a single set of control parameters by hand that work well across the broad array of environments and conditions that a robot might encounter. Automated adaptation approaches must ...
Hersh Sanghvi;Spencer Folk;Camillo Jose Taylor
University of Pennsylvania;University of Pennsylvania;University of Pennsylvania
Poster
main
Controller Adaptation;Robot Model Learning;Meta-Learning
https://openreview.net/forum?id=xeFKtSXPMd
2
OCCAM: Online Continuous Controller Adaptation with Meta-Learned Models Control tuning and adaptation present a significant challenge to the usage of robots in diverse environments. It is often nontrivial to find a single set of control parameters by hand that work well across the broad array of environments and condit...
[ -0.03190053254365921, -0.011291059665381908, -0.03804391995072365, 0.015940872952342033, -0.02301422320306301, -0.03764938935637474, -0.009496890008449554, 0.019087715074419975, 0.04850834608078003, 0.03723607212305069, -0.05399418622255325, -0.011713770218193531, 0.018721366301178932, 0.0...
corl_2024_y8XkuQIrvI
y8XkuQIrvI
corl
2,024
MILES: Making Imitation Learning Easy with Self-Supervision
Data collection in imitation learning often requires significant, laborious human supervision, such as numerous demonstrations, and/or frequent environment resets for methods that incorporate reinforcement learning. In this work, we propose an alternative approach, MILES: a fully autonomous, self-supervised data collec...
Georgios Papagiannis;Edward Johns
Imperial College London;Imperial College London
Poster
main
Imitation Learning;Robotic Manipulation;Self-Supervised Data Collection
https://openreview.net/forum?id=y8XkuQIrvI
3
MILES: Making Imitation Learning Easy with Self-Supervision Data collection in imitation learning often requires significant, laborious human supervision, such as numerous demonstrations, and/or frequent environment resets for methods that incorporate reinforcement learning. In this work, we propose an alternative appr...
[ -0.04843391478061676, -0.001900124829262495, -0.04391191899776459, 0.020591886714100838, -0.029486387968063354, -0.04398666322231293, 0.010426744818687439, 0.02901923842728138, 0.014490933157503605, 0.035372454673051834, -0.03957678750157356, -0.013220290653407574, 0.0335412323474884, 0.03...
corl_2024_yNQu9zqx6X
yNQu9zqx6X
corl
2,024
Robust Manipulation Primitive Learning via Domain Contraction
Contact-rich manipulation plays an important role in everyday life, but uncertain parameters pose significant challenges to model-based planning and control. To address this issue, domain adaptation and domain randomization have been proposed to learn robust policies. However, they either lose the generalization abilit...
Teng Xue;Amirreza Razmjoo;Suhan Shetty;Sylvain Calinon
Idiap Research Institute;;EPFL - EPF Lausanne;EPFL - EPF Lausanne
Poster
main
Robust policy learning;Contact-rich manipulation;Sim-to-real
https://openreview.net/forum?id=yNQu9zqx6X
3
Robust Manipulation Primitive Learning via Domain Contraction Contact-rich manipulation plays an important role in everyday life, but uncertain parameters pose significant challenges to model-based planning and control. To address this issue, domain adaptation and domain randomization have been proposed to learn robust...
[ -0.023700308054685593, -0.0018301487434655428, -0.009256464429199696, -0.011981718242168427, -0.03542829677462578, -0.004179505165666342, 0.03150017186999321, -0.007940825074911118, 0.03488324582576752, 0.04157421365380287, -0.07198428362607956, 0.008852374739944935, -0.01369204930961132, ...
corl_2024_ySI0tBYxpz
ySI0tBYxpz
corl
2,024
Gaitor: Learning a Unified Representation Across Gaits for Real-World Quadruped Locomotion
The current state-of-the-art in quadruped locomotion is able to produce a variety of complex motions. These methods either rely on switching between a discrete set of skills or learn a distribution across gaits using complex black-box models. Alternatively, we present Gaitor, which learns a disentangled and 2D represen...
Alexander Luis Mitchell;Wolfgang Merkt;Aristotelis Papatheodorou;Ioannis Havoutis;Ingmar Posner
University of Oxford;University of Oxford, University of Oxford;University of Oxford;;University of Oxford
Poster
main
Representation Learning;Learning for Control;Quadruped Control
https://openreview.net/forum?id=ySI0tBYxpz
3
Gaitor: Learning a Unified Representation Across Gaits for Real-World Quadruped Locomotion The current state-of-the-art in quadruped locomotion is able to produce a variety of complex motions. These methods either rely on switching between a discrete set of skills or learn a distribution across gaits using complex blac...
[ -0.06510942429304123, -0.011264670640230179, -0.014649620279669762, 0.03327609598636627, -0.009720170870423317, -0.030723510310053825, -0.03958357125520706, 0.009655431844294071, 0.007375677116215229, 0.0361616276204586, -0.003775699296966195, -0.025951284915208817, -0.019551323726773262, ...
corl_2024_yYujuPxjDK
yYujuPxjDK
corl
2,024
Language-guided Manipulator Motion Planning with Bounded Task Space
Language-based robot control is a powerful and versatile method to control a robot manipulator where large language models (LLMs) are used to reason about the environment. However, the generated robot motions by these controllers often lack safety and performance, resulting in jerky movements. In this work, a novel mod...
Thies Oelerich;Christian Hartl-Nesic;Andreas Kugi
Technische Universität Wien;;Technische Universität Wien
Poster
main
Vision Language Models;Manipulation Planning;Path-following MPC
https://openreview.net/forum?id=yYujuPxjDK
3
Language-guided Manipulator Motion Planning with Bounded Task Space Language-based robot control is a powerful and versatile method to control a robot manipulator where large language models (LLMs) are used to reason about the environment. However, the generated robot motions by these controllers often lack safety and ...
[ -0.06134950742125511, -0.011651297099888325, -0.007075631059706211, 0.0015807260060682893, -0.007714307866990566, 0.024306228384375572, -0.009370306506752968, -0.001140494947321713, 0.0012841973220929503, 0.021331818774342537, -0.04229867830872536, -0.0009682802483439445, 0.01230822224169969...
corl_2024_ylZHvlwUcI
ylZHvlwUcI
corl
2,024
Theia: Distilling Diverse Vision Foundation Models for Robot Learning
Vision-based robot policy learning, which maps visual inputs to actions, necessitates a holistic understanding of diverse visual tasks beyond single-task needs like classification or segmentation. Inspired by this, we introduce Theia, a vision foundation model for robot learning that distills multiple off-the-shelf vis...
Jinghuan Shang;Karl Schmeckpeper;Brandon B. May;Maria Vittoria Minniti;Tarik Kelestemur;David Watkins;Laura Herlant
Department of Computer Science, State University of New York, Stony Brook;The Robotics and AI Institute;;The AI Institute;Boston Dynamics AI Institute;;The Robotics and AI Institute
Poster
main
visual representation;robot learning;distillation;foundation model
https://github.com/bdaiinstitute/theia
https://openreview.net/forum?id=ylZHvlwUcI
18
Theia: Distilling Diverse Vision Foundation Models for Robot Learning Vision-based robot policy learning, which maps visual inputs to actions, necessitates a holistic understanding of diverse visual tasks beyond single-task needs like classification or segmentation. Inspired by this, we introduce Theia, a vision founda...
[ -0.03517217934131622, -0.043021444231271744, -0.015044423751533031, 0.02455698326230049, -0.03134099021553993, -0.01602558046579361, -0.02992064878344536, 0.03453676402568817, 0.027285538613796234, 0.03545251116156578, -0.020893994718790054, -0.021193014457821846, -0.008078200742602348, 0....
corl_2024_ypaYtV1CoG
ypaYtV1CoG
corl
2,024
Vocal Sandbox: Continual Learning and Adaptation for Situated Human-Robot Collaboration
We introduce Vocal Sandbox, a framework for enabling seamless human-robot collaboration in situated environments. Systems in our framework are characterized by their ability to *adapt and continually learn* at multiple levels of abstraction from diverse teaching modalities such as spoken dialogue, object keypoints, and...
Jennifer Grannen;Siddharth Karamcheti;Suvir Mirchandani;Percy Liang;Dorsa Sadigh
Computer Science Department, Stanford University;Stanford University;Stanford University;Stanford University;Stanford University
Poster
main
Continual Learning;Multimodal Teaching;Human-Robot Interaction
https://openreview.net/forum?id=ypaYtV1CoG
0
Vocal Sandbox: Continual Learning and Adaptation for Situated Human-Robot Collaboration We introduce Vocal Sandbox, a framework for enabling seamless human-robot collaboration in situated environments. Systems in our framework are characterized by their ability to *adapt and continually learn* at multiple levels of abs...
[ -0.03546343743801117, -0.00456432718783617, -0.0018994978163391352, 0.0060073924250900745, -0.007206104230135679, 0.000254005950409919, 0.011562962085008621, 0.010668538510799408, -0.015657024458050728, -0.0020101480185985565, 0.0032826270908117294, -0.04968200996518135, 0.020138364285230637...
corl_2024_yqLFb0RnDW
yqLFb0RnDW
corl
2,024
Unpacking Failure Modes of Generative Policies: Runtime Monitoring of Consistency and Progress
Robot behavior policies trained via imitation learning are prone to failure under conditions that deviate from their training data. Thus, algorithms that monitor learned policies at test time and provide early warnings of failure are necessary to facilitate scalable deployment. We propose Sentinel, a runtime monitoring...
Christopher Agia;Rohan Sinha;Jingyun Yang;Ziang Cao;Rika Antonova;Marco Pavone;Jeannette Bohg
Stanford University;Stanford University;Stanford University;;;Stanford University;Stanford University
Poster
main
Failure Detection;Generative Policies;Vision Language Models
https://openreview.net/forum?id=yqLFb0RnDW
9
Unpacking Failure Modes of Generative Policies: Runtime Monitoring of Consistency and Progress Robot behavior policies trained via imitation learning are prone to failure under conditions that deviate from their training data. Thus, algorithms that monitor learned policies at test time and provide early warnings of fai...
[ -0.06809806823730469, -0.020983275026082993, 0.01893884502351284, 0.006063590291887522, -0.02906806208193302, 0.01701522432267666, -0.005603593774139881, 0.03397469222545624, 0.0009856238029897213, 0.034866806119680405, -0.008549429476261139, -0.025146475061774254, -0.008428622968494892, 0...
corl_2024_zIWu9Kmlqk
zIWu9Kmlqk
corl
2,024
LeLaN: Learning A Language-Conditioned Navigation Policy from In-the-Wild Video
We present our method, LeLaN, which uses action-free egocentric data to learn robust language-conditioned object navigation. By leveraging the knowledge of large vision and language models and grounding this knowledge using pre-trained segmentation and depth estimation models, we can label in-the-wild data from a varie...
Noriaki Hirose;Catherine Glossop;Ajay Sridhar;Oier Mees;Sergey Levine
Toyota Central R&D Labs., Inc;University of California, Berkeley;University of California, Berkeley;Electrical Engineering & Computer Science Department, University of California, Berkeley;Google
Poster
main
Language-conditioned navigation policy;data augmentation
https://github.com/NHirose/learning-language-navigation
https://openreview.net/forum?id=zIWu9Kmlqk
7
LeLaN: Learning A Language-Conditioned Navigation Policy from In-the-Wild Video We present our method, LeLaN, which uses action-free egocentric data to learn robust language-conditioned object navigation. By leveraging the knowledge of large vision and language models and grounding this knowledge using pre-trained segm...
[ -0.06475963443517685, -0.025210922583937645, -0.03402000293135643, 0.00675424188375473, -0.015766041353344917, 0.00864321831613779, -0.014715585857629776, 0.012384312227368355, -0.03405686095356941, 0.021432969719171524, -0.019700640812516212, -0.008928867988288403, 0.006454769987612963, -...
corl_2024_zeYaLS2tw5
zeYaLS2tw5
corl
2,024
Sparse Diffusion Policy: A Sparse, Reusable, and Flexible Policy for Robot Learning
The increasing complexity of tasks in robotics demands efficient strategies for multitask and continual learning. Traditional models typically rely on a universal policy for all tasks, facing challenges such as high computational costs and catastrophic forgetting when learning new tasks. To address these issues, we int...
Yixiao Wang;Yifei Zhang;Mingxiao Huo;Thomas Tian;Xiang Zhang;Yichen Xie;Chenfeng Xu;Pengliang Ji;Wei Zhan;Mingyu Ding;Masayoshi Tomizuka
University of California, Berkeley;University of Chinese Academy of Sciences;Carnegie Mellon University;University of California, Berkeley;University of California, Berkeley;Waymo;University of California, Berkeley;;;University of California, Berkeley;
Poster
main
Robot Policy;Multitask;Continual learning;Mixture of Experts
https://github.com/AnthonyHuo/SDP
https://openreview.net/forum?id=zeYaLS2tw5
17
Sparse Diffusion Policy: A Sparse, Reusable, and Flexible Policy for Robot Learning The increasing complexity of tasks in robotics demands efficient strategies for multitask and continual learning. Traditional models typically rely on a universal policy for all tasks, facing challenges such as high computational costs ...
[ -0.04772655665874481, -0.020282862707972527, -0.026613183319568634, 0.017311489209532738, -0.016074953600764275, -0.02838493511080742, 0.03667156398296356, 0.011534840799868107, 0.04034425690770149, 0.02947382442653179, -0.02526591531932354, 0.012402260676026344, -0.008074387907981873, -0....
corl_2024_zr2GPi3DSb
zr2GPi3DSb
corl
2,024
Gentle Manipulation of Tree Branches: A Contact-Aware Policy Learning Approach
Learning to interact with deformable tree branches with minimal damage is challenging due to their intricate geometry and inscrutable dynamics. Furthermore, traditional vision-based modelling systems suffer from implicit occlusions in dense foliage, severely changing lighting conditions, and limited field of view, in a...
Jay Jacob;Shizhe Cai;Paulo Vinicius Koerich Borges;Tirthankar Bandyopadhyay;Fabio Ramos
;University of Sydney;CSIRO;, CSIRO;NVIDIA
Poster
main
Reinforcement Learning;Sim-to-Real;Deformable Manipulation
https://openreview.net/forum?id=zr2GPi3DSb
3
Gentle Manipulation of Tree Branches: A Contact-Aware Policy Learning Approach Learning to interact with deformable tree branches with minimal damage is challenging due to their intricate geometry and inscrutable dynamics. Furthermore, traditional vision-based modelling systems suffer from implicit occlusions in dense ...
[ -0.04792873561382294, 0.0014609048375859857, 0.008461265824735165, -0.026360806077718735, -0.02737468294799328, -0.0665472075343132, 0.040370743721723557, 0.012479905970394611, 0.040223270654678345, 0.042545970529317856, -0.01941113919019699, -0.016498547047376633, 0.0033181433100253344, 0...
corl_2025_06YyNxzwae
06YyNxzwae
corl
2,025
Fail2Progress: Learning from Real-World Robot Failures with Stein Variational Inference
Skill effect models for long-horizon manipulation tasks are prone to failures in conditions not covered by training data distributions. Therefore, enabling robots to reason about and learn from failures is necessary. We investigate the problem of efficiently generating a dataset targeted to observed failures. After fin...
Yixuan Huang;Novella Alvina;Mohanraj Devendran Shanthi;Tucker Hermans
Princeton University+University of Utah;University of Utah;, University of Utah;NVIDIA+University of Utah
Poster
main
Learning from failures;Variational inference;Skill effect models
https://openreview.net/forum?id=06YyNxzwae
-1
Fail2Progress: Learning from Real-World Robot Failures with Stein Variational Inference Skill effect models for long-horizon manipulation tasks are prone to failures in conditions not covered by training data distributions. Therefore, enabling robots to reason about and learn from failures is necessary. We investigate ...
[ -0.036798857152462006, -0.0051994360983371735, -0.019149480387568474, 0.02089034393429756, -0.00618561590090394, -0.0018600831972435117, -0.0039724987000226974, 0.03333565220236778, -0.01612149178981781, 0.024649864062666893, -0.03533579409122467, -0.0013727807672694325, -0.01470472756773233...
corl_2025_0ViTEgiFiQ
0ViTEgiFiQ
corl
2,025
Disentangled Multi-Context Meta-Learning: Unlocking Robust and Generalized Task Learning
In meta-learning and its downstream tasks, many methods use implicit adaptation to represent task-specific variations. However, implicit approaches hinder interpretability and make it difficult to understand which task factors drive performance. In this work, we introduce a disentangled multi-context meta-learning fram...
Seonsoo Kim;Jun-Gill Kang;Taehong Kim;Seongil Hong
Agency for Defense Development;Agency For Defense Development;Agency for defense development;Agency for Defense Development
Poster
main
Meta-Learning;Multi Task Learning;Quadruped Robot Locomotion
https://openreview.net/forum?id=0ViTEgiFiQ
-1
Disentangled Multi-Context Meta-Learning: Unlocking Robust and Generalized Task Learning In meta-learning and its downstream tasks, many methods use implicit adaptation to represent task-specific variations. However, implicit approaches hinder interpretability and make it difficult to understand which task factors driv...
[ -0.04264608025550842, -0.03201206400990486, -0.009497278369963169, 0.02128637209534645, -0.022404776886105537, -0.012861661612987518, 0.0006067578215152025, 0.014282586984336376, 0.030893657356500626, 0.014291753992438316, -0.0522533655166626, -0.02434823475778103, -0.005078660324215889, 0...
corl_2025_19LSN4QnV4
19LSN4QnV4
corl
2,025
FOMO-3D: Using Vision Foundation Models for Long-Tailed 3D Object Detection
In order to navigate complex traffic environments, self-driving vehicles must recognize many semantic classes pertaining to vulnerable road users or traffic control devices. However, many safety-critical objects (e.g., construction worker) appear infrequently in nominal traffic conditions, leading to a severe shortage...
Anqi Joyce Yang;James Tu;Nikita Dvornik;Enxu Li;Raquel Urtasun
University of Toronto+Waabi Innovation Inc;Department of Computer Science, University of Toronto;Palona AI;Department of Computer Science, University of Toronto+Waabi;Waabi+Department of Computer Science, University of Toronto
Poster
main
Long-Tailed 3D Object Detection;Vision Foundation Model;Multimodal Fusion;Autonomous Vehicles
https://openreview.net/forum?id=19LSN4QnV4
-1
FOMO-3D: Using Vision Foundation Models for Long-Tailed 3D Object Detection In order to navigate complex traffic environments, self-driving vehicles must recognize many semantic classes pertaining to vulnerable road users or traffic control devices. However, many safety-critical objects (e.g., construction worker) app...
[ -0.03752383962273598, -0.050093308091163635, -0.000026820869607036002, 0.010077725164592266, -0.021189063787460327, 0.016242487356066704, 0.012403354048728943, 0.03975718095898628, -0.032558806240558624, -0.0029347222298383713, -0.03126678988337517, -0.011028280481696129, 0.02624638378620147...
corl_2025_1D6XYy6ofW
1D6XYy6ofW
corl
2,025
LaVA-Man: Learning Visual Action Representations for Robot Manipulation
Visual-textual understanding is essential for language-guided robot manipulation. Recent works leverage pre-trained vision-language models to measure the similarity between encoded visual observations and textual instructions, and then train a model to map this similarity to robot actions. However, this two-step approa...
Chaoran Zhu;Hengyi Wang;Yik Lung Pang;Changjae Oh
Queen Mary University of London;University College London;;
Poster
main
Robot manipulation;self-supervised representation learning
https://openreview.net/forum?id=1D6XYy6ofW
-1
LaVA-Man: Learning Visual Action Representations for Robot Manipulation Visual-textual understanding is essential for language-guided robot manipulation. Recent works leverage pre-trained vision-language models to measure the similarity between encoded visual observations and textual instructions, and then train a mode...
[ -0.023110290989279747, -0.022999538108706474, 0.013447161763906479, -0.000004346537025412545, -0.007618827745318413, 0.014591601677238941, 0.05076142027974129, 0.022944163531064987, 0.026155976578593254, 0.027208121493458748, -0.02019381709396839, -0.057886481285095215, 0.016049839556217194,...
corl_2025_1HW2UhshIT
1HW2UhshIT
corl
2,025
Force-Modulated Visual Policy for Robot-Assisted Dressing with Arm Motions
Robot-assisted dressing has the potential to significantly improve the lives of individuals with mobility impairments. To ensure an effective and comfortable dressing experience, the robot must be able to handle challenging deformable garments, apply appropriate forces, and adapt to limb movements throughout the dressi...
Alexis Yihong Hao;Yufei Wang;Navin Sriram Ravie;Bharath Hegde;David Held;Zackory Erickson
Carnegie Mellon University;School of Computer Science, Carnegie Mellon University;Indian Institute of Technology Madras;Carnegie Mellon University;Carnegie Mellon University;Carnegie Mellon University
Poster
main
Robot-Assisted Dressing;Multi-Modal Learning;Physical Human Robot Interaction;Deformable Object Manipulation
https://openreview.net/forum?id=1HW2UhshIT
-1
Force-Modulated Visual Policy for Robot-Assisted Dressing with Arm Motions Robot-assisted dressing has the potential to significantly improve the lives of individuals with mobility impairments. To ensure an effective and comfortable dressing experience, the robot must be able to handle challenging deformable garments, ...
[ -0.06051256135106087, -0.03103354200720787, -0.026692258194088936, 0.038218460977077484, -0.03251223266124725, -0.029289444908499718, 0.005706228315830231, 0.042389124631881714, 0.024417350068688393, 0.05164041742682457, -0.04110001027584076, 0.01688171550631523, 0.012644698843359947, 0.01...
corl_2025_1K3kjo91Q1
1K3kjo91Q1
corl
2,025
Learning from 10 Demos: Generalisable and Sample-Efficient Policy Learning with Oriented Affordance Frames
Imitation learning has unlocked the potential for robots to exhibit highly dexterous behaviours. However, it still struggles with long-horizon, multi-object tasks due to poor sample efficiency and limited generalisation. Existing methods require a substantial number of demonstrations to cover possible task variations, ...
Krishan Rana;Jad Abou-Chakra;Sourav Garg;Robert Lee;Ian Reid;Niko Suenderhauf
Queensland University of Technology;The AI Institute+Queensland University of Technology;University of Adelaide;Woven By Toyota, Inc.;Mohamed bin Zayed University of Artificial Intelligence+University of Adelaide;Queensland University of Technology
Poster
main
behaviour cloning;spatial generalisation;intra-category generalisation;long-horizon tasks;affordances
https://openreview.net/forum?id=1K3kjo91Q1
-1
Learning from 10 Demos: Generalisable and Sample-Efficient Policy Learning with Oriented Affordance Frames Imitation learning has unlocked the potential for robots to exhibit highly dexterous behaviours. However, it still struggles with long-horizon, multi-object tasks due to poor sample efficiency and limited generali...
[ -0.021124886348843575, -0.04565218463540077, -0.0002479953982401639, 0.01805897243320942, -0.03671617433428764, -0.04860593006014824, -0.0003785654262173921, -0.0019956473261117935, 0.017002727836370468, 0.022975649684667587, -0.04281060770153999, 0.006304750218987465, 0.00445164879783988, ...
corl_2025_1NBdplgILy
1NBdplgILy
corl
2,025
Vision in Action: Learning Active Perception from Human Demonstrations
We present Vision in Action (ViA), an active perception system for bimanual robot manipulation. ViA learns task-relevant active perceptual strategies (e.g., searching, tracking, and focusing) directly from human demonstrations. On the hardware side, ViA employs a simple yet effective 6-DoF robotic neck to enable flexib...
Haoyu Xiong;Xiaomeng Xu;Jimmy Wu;Yifan Hou;Jeannette Bohg;Shuran Song
Massachusetts Institute of Technology;Stanford University;Princeton University;Stanford University;Stanford University;Stanford University
Poster
main
Active Perception;Bimanual Manipulation;Imitation Learning;Teleoperation Systems
https://openreview.net/forum?id=1NBdplgILy
-1
Vision in Action: Learning Active Perception from Human Demonstrations We present Vision in Action (ViA), an active perception system for bimanual robot manipulation. ViA learns task-relevant active perceptual strategies (e.g., searching, tracking, and focusing) directly from human demonstrations. On the hardware side,...
[ -0.036072999238967896, -0.04335605353116989, -0.019177448004484177, 0.017751261591911316, -0.012531421147286892, 0.005799822974950075, 0.03888733685016632, 0.04221510514616966, 0.04350817948579788, 0.034970082342624664, -0.01423333678394556, -0.049212925136089325, 0.025176936760544777, 0.0...
corl_2025_1TdRe3wPqK
1TdRe3wPqK
corl
2,025
Adapting by Analogy: OOD Generalization of Visuomotor Policies via Functional Correspondence
End-to-end visuomotor policies trained using behavior cloning have shown a remarkable ability to generate complex, multi-modal low-level robot behaviors. However, at deployment time, these policies still struggle to act reliably when faced with out-of-distribution (OOD) visuals induced by objects, backgrounds, or envir...
Pranay Gupta;Henny Admoni;Andrea Bajcsy
Carnegie Mellon University;Carnegie Mellon University;
Poster
main
Visuomotor Policy;Out-of-Distribution Generalization;Functional Correspondence;Deployment-Time Adaptation
https://openreview.net/forum?id=1TdRe3wPqK
-1
Adapting by Analogy: OOD Generalization of Visuomotor Policies via Functional Correspondence End-to-end visuomotor policies trained using behavior cloning have shown a remarkable ability to generate complex, multi-modal low-level robot behaviors. However, at deployment time, these policies still struggle to act reliabl...
[ -0.052557650953531265, -0.011976796202361584, -0.030689386650919914, -0.006518957205116749, -0.05322200432419777, -0.0071556284092366695, -0.009208661504089832, -0.01802978478372097, 0.02786588855087757, 0.017070164903998375, -0.02074255608022213, -0.03989804908633232, 0.03786808252334595, ...
corl_2025_1cA6OYsfoJ
1cA6OYsfoJ
corl
2,025
From Real World to Logic and Back: Learning Generalizable Relational Concepts For Long Horizon Robot Planning
Humans efficiently generalize from limited demonstrations, but robots still struggle to transfer learned knowledge to complex, unseen tasks with longer horizons and increased complexity. We propose the first known method enabling robots to autonomously invent relational concepts directly from small sets of unannotated...
Naman Shah;Jayesh Nagpal;Siddharth Srivastava
Allen Institute for Artificial Intelligence+Brown University;;
Poster
main
Learnng symbolic abstractions;Symbolic world model learning;Learning for task and motion planning;learning for planning
https://openreview.net/forum?id=1cA6OYsfoJ
-1
From Real World to Logic and Back: Learning Generalizable Relational Concepts For Long Horizon Robot Planning Humans efficiently generalize from limited demonstrations, but robots still struggle to transfer learned knowledge to complex, unseen tasks with longer horizons and increased complexity. We propose the first k...
[ -0.03369223698973656, 0.01002421136945486, 0.02102816477417946, 0.015929877758026123, -0.022388918325304985, -0.03314793482422829, -0.01761721260845661, -0.001338073518127203, 0.003270342480391264, -0.017136413604021072, -0.043761808425188065, -0.012283061631023884, 0.0005136841791681945, ...
corl_2025_1n1Liq6So4
1n1Liq6So4
corl
2,025
Meta-Optimization and Program Search using Language Models for Task and Motion Planning
Intelligent interaction with the real world requires robotic agents to jointly reason over high-level plans and low-level controls. This requirement is formalized in the task and motion planning (TAMP) problem, in which symbolic planning and continuous trajectory generation must be solved in a coordinated manner. Recen...
Denis Shcherba;Eckart Cobo-Briesewitz;Cornelius V. Braun;Marc Toussaint
Technische Universität Berlin;Technische Universität Berlin;Technische Universität Berlin;TU Berlin
Poster
main
Task and Motion Planning;LLMs as Optimizers;Trajectory Optimization
https://openreview.net/forum?id=1n1Liq6So4
-1
Meta-Optimization and Program Search using Language Models for Task and Motion Planning Intelligent interaction with the real world requires robotic agents to jointly reason over high-level plans and low-level controls. This requirement is formalized in the task and motion planning (TAMP) problem, in which symbolic pla...
[ -0.021518705412745476, -0.049046121537685394, 0.010806295089423656, 0.008036654442548752, -0.019340546801686287, 0.019284214824438095, 0.009961320087313652, -0.012486857362091541, -0.008468530140817165, 0.047581497579813004, -0.03605227917432785, -0.03595839440822601, -0.0009206710383296013,...
corl_2025_1otaE496Vm
1otaE496Vm
corl
2,025
CaRL: Learning Scalable Planning Policies with Simple Rewards
We investigate reinforcement learning (RL) for privileged planning in autonomous driving. State-of-the-art approaches for this task are rule-based, but these methods do not scale to the long tail. RL, on the other hand, is scalable and does not suffer from compounding errors like imitation learning. Contemporary RL ap...
Bernhard Jaeger;Daniel Dauner;Jens Beißwenger;Simon Gerstenecker;Kashyap Chitta;Andreas Geiger
Eberhard-Karls-Universität Tübingen;Eberhard-Karls-Universität Tübingen;Max-Planck-Institute for Intelligent Systems, Max-Planck Institute+Eberhard-Karls-Universität Tübingen;;NVIDIA+University of Tübingen;University of Tuebingen
Poster
main
Autonomous Driving;Reinforcement Learning;Planning
https://openreview.net/forum?id=1otaE496Vm
-1
CaRL: Learning Scalable Planning Policies with Simple Rewards We investigate reinforcement learning (RL) for privileged planning in autonomous driving. State-of-the-art approaches for this task are rule-based, but these methods do not scale to the long tail. RL, on the other hand, is scalable and does not suffer from ...
[ -0.08328351378440857, -0.017161395400762558, 0.014619311317801476, -0.0020590415224432945, -0.02663135714828968, 0.021994149312376976, 0.036911431699991226, 0.03814057260751724, -0.012617303989827633, 0.007002370897680521, -0.014544817619025707, 0.0022627341095358133, 0.006299340166151524, ...
corl_2025_23FdMTxEh7
23FdMTxEh7
corl
2,025
Mastering Multi-Drone Volleyball through Hierarchical Co-Self-Play Reinforcement Learning
In this paper, we tackle the problem of learning to play 3v3 multi-drone volleyball, a new embodied competitive task that requires both high-level strategic coordination and low-level agile control. The task is turn-based, multi-agent, and physically grounded, posing significant challenges due to its long-horizon depen...
Ruize Zhang;Sirui Xiang;Zelai Xu;Feng Gao;Shilong Ji;Wenhao Tang;Wenbo Ding;Chao Yu;Yu Wang
Tsinghua University;Tsinghua University;Tsinghua University;IIIS, Tsinghua University;Tsinghua University;Tsinghua University;Tsinghua Univeresity;Tsinghua University;Tsinghua University
Poster
main
Multi-Agent;Reinforcement Learning;Self-play;Drone Volleyball
https://openreview.net/forum?id=23FdMTxEh7
-1
Mastering Multi-Drone Volleyball through Hierarchical Co-Self-Play Reinforcement Learning In this paper, we tackle the problem of learning to play 3v3 multi-drone volleyball, a new embodied competitive task that requires both high-level strategic coordination and low-level agile control. The task is turn-based, multi-a...
[ -0.024593550711870193, -0.027402129024267197, -0.01934594102203846, 0.0022692678030580282, -0.024094657972455025, 0.01712864078581333, -0.00035800153273157775, 0.03954184427857399, 0.007007590960711241, -0.0047902911901474, -0.021840402856469154, 0.018708467483520508, -0.026552164927124977, ...
corl_2025_2CIKnIwSta
2CIKnIwSta
corl
2,025
Rapid Mismatch Estimation via Neural Network Informed Variational Inference
With robots increasingly operating in human-centric environments, ensuring soft and safe physical interactions, whether with humans, surroundings, or other machines, is essential. While compliant hardware can facilitate such interactions, this work focuses on impedance controllers that allow torque-controlled robots to...
Mateusz Jaszczuk;Nadia Figueroa
University of Pennsylvania;University of Pennsylvania
Poster
main
Passive Impedance Control;Learning Residual Inverse Dynamics;Model Mismatch Estimation
https://openreview.net/forum?id=2CIKnIwSta
-1
Rapid Mismatch Estimation via Neural Network Informed Variational Inference With robots increasingly operating in human-centric environments, ensuring soft and safe physical interactions, whether with humans, surroundings, or other machines, is essential. While compliant hardware can facilitate such interactions, this ...
[ -0.009601066820323467, -0.012415651232004166, 0.003830705303698778, 0.002884023590013385, -0.02253519557416439, 0.003907088190317154, 0.03525637835264206, 0.002546088071539998, 0.03314543887972832, 0.019368786364793777, -0.026942241936922073, -0.010767638683319092, -0.009582550264894962, 0...
corl_2025_2dXMfk3qRU
2dXMfk3qRU
corl
2,025
First Order Model-Based RL through Decoupled Backpropagation
There is growing interest in reinforcement learning (RL) methods that leverage the simulator's derivatives to improve learning efficiency. While early gradient-based approaches have demonstrated superior performance compared to derivative-free methods, accessing simulator gradients is often impractical due to their imp...
Joseph Amigo;Rooholla Khorrambakht;Elliot Chane-Sane;Nicolas Mansard;Ludovic Righetti
New York University;New York University;LAAS / CNRS;LAAS / CNRS;New York University+Max-Planck Institute
Poster
main
Model-Based Reinforcement Learning;Quadruped Locomotion;Sim-to-Real Transfer
https://openreview.net/forum?id=2dXMfk3qRU
-1
First Order Model-Based RL through Decoupled Backpropagation There is growing interest in reinforcement learning (RL) methods that leverage the simulator's derivatives to improve learning efficiency. While early gradient-based approaches have demonstrated superior performance compared to derivative-free methods, access...
[ -0.058004818856716156, -0.039834633469581604, 0.0002593400713521987, -0.024441471323370934, -0.03702082857489586, -0.010206934995949268, 0.04160015657544136, 0.014501204714179039, 0.010804638266563416, 0.0005566112813539803, -0.06116804853081703, -0.02916792593896389, 0.013232233934104443, ...
corl_2025_2xvxn3Hm3n
2xvxn3Hm3n
corl
2,025
NeuralSVCD for Efficient Swept Volume Collision Detection
Robot manipulation in unstructured environments requires efficient and reliable Swept Volume Collision Detection (SVCD) for safe motion planning. Traditional discrete methods potentially miss collisions between these points, whereas SVCD continuously checks for collisions along the entire trajectory. Existing SVCD meth...
Hojin Jung;Dongwon Son;Beomjoon Kim
Korea Advanced Institute of Science & Technology;KAIST;Korea Advanced Institute of Science & Technology
Poster
main
Neural swept-volume collision detection;Motion planning
https://openreview.net/forum?id=2xvxn3Hm3n
-1
NeuralSVCD for Efficient Swept Volume Collision Detection Robot manipulation in unstructured environments requires efficient and reliable Swept Volume Collision Detection (SVCD) for safe motion planning. Traditional discrete methods potentially miss collisions between these points, whereas SVCD continuously checks for ...
[ -0.06709452718496323, -0.018493296578526497, -0.01613910309970379, 0.004165819846093655, 0.0019760055001825094, 0.01567929983139038, 0.009536325000226498, -0.018815159797668457, 0.057420261204242706, 0.035037025809288025, 0.016332220286130905, -0.02179468609392643, -0.004149726592004299, -...
corl_2025_2y7TSgwqAB
2y7TSgwqAB
corl
2,025
GENNAV: Polygon Mask Generation for Generalized Referring Navigable Regions
We focus on the task of identifying the location of target regions from a natural language instruction and a front camera image captured by a mobility. This task is challenging because it requires both existence prediction and segmentation mask generation, particularly for stuff-type target regions with ambiguous bound...
Kei Katsumata;Yui Iioka;Naoki Hosomi;Teruhisa Misu;Kentaro Yamada;Komei Sugiura
Keio University;Keio University, Tokyo Institute of Technology;Honda R&D Co., Ltd.;Honda Research Institute USA, Inc.;Honda;Keio University
Poster
main
Autonomous driving;Vision and Language;Semantic Understanding
https://openreview.net/forum?id=2y7TSgwqAB
-1
GENNAV: Polygon Mask Generation for Generalized Referring Navigable Regions We focus on the task of identifying the location of target regions from a natural language instruction and a front camera image captured by a mobility. This task is challenging because it requires both existence prediction and segmentation mask...
[ -0.07341090589761734, -0.04259323328733444, 0.006497983355075121, -0.002005291637033224, -0.01719753071665764, 0.03266226500272751, 0.0020483785774558783, 0.018203670158982277, 0.0019668627064675093, 0.03569931536912918, -0.02701670490205288, 0.04106539115309715, -0.01715094968676567, 0.01...
corl_2025_3CnxNqmklv
3CnxNqmklv
corl
2,025
DreamGen: Unlocking Generalization in Robot Learning through Video World Models
In this work, we unlock new capabilities in robot learning from neural trajectories, synthetic robot data generated from video world models. Our proposed recipe is simple, but powerful: we take the most recent state-of-the-art video generative models (world models), adapt them to the target robot embodiment, and genera...
Joel Jang;Seonghyeon Ye;Zongyu Lin;Jiannan Xiang;Johan Bjorck;Yu Fang;Fengyuan Hu;Spencer Huang;Kaushil Kundalia;Yen-Chen Lin;Loïc Magne;Ajay Mandlekar;Avnish Narayan;You Liang Tan;Guanzhi Wang;Jing Wang;Qi Wang;Yinzhen Xu;Xiaohui Zeng;Kaiyuan Zheng;Ruijie Zheng;Ming-Yu Liu;Luke Zettlemoyer;Dieter Fox;Jan Kautz;Scott R...
Department of Computer Science, University of Washington;Korea Advanced Institute of Science & Technology;University of California, Los Angeles;University of California, San Diego;Microsoft;;;;NVIDIA;Massachusetts Institute of Technology;;NVIDIA;;NVIDIA;California Institute of Technology;Nanyang Technological Universit...
Poster
main
Video World Models;Synthetic Data;Behavior Generalization;Environment Generalization
https://openreview.net/forum?id=3CnxNqmklv
-1
DreamGen: Unlocking Generalization in Robot Learning through Video World Models In this work, we unlock new capabilities in robot learning from neural trajectories, synthetic robot data generated from video world models. Our proposed recipe is simple, but powerful: we take the most recent state-of-the-art video generat...
[ -0.045898422598838806, -0.022259827703237534, -0.0071750301867723465, -0.00020664495241362602, -0.028845254331827164, 0.000611714378464967, -0.053300224244594574, 0.034759439527988434, -0.012318192981183529, -0.012154918164014816, -0.030006321147084236, -0.02313062734901905, 0.00819549895823...
corl_2025_3p7rTnLJM8
3p7rTnLJM8
corl
2,025
Lucid-XR: An Extended-Reality Data Engine for Robotic Manipulation
We introduce Lucid-XR, a generative data engine for creating diverse and realistic-looking data to train real-world robot systems. At the core of Lucid-XR is vuer, a web-based physics simulation environment that runs directly on the XR headset, enabling internet-scale access to immersive, latency-free virtual interacti...
Yajvan Ravan;Adam Rashid;Alan Yu;Kai McClennen;Gio Huh;Kevin Yang;Zhutian Yang;Qinxi Yu;Xiaolong Wang;Phillip Isola;Ge Yang
Massachusetts Institute of Technology;Massachusetts Institute of Technology;Massachusetts Institute of Technology;Massachusetts Institute of Technology;California Institute of Technology;;Google;University of Illinois, Urbana Champaign;University of California, San Diego;Massachusetts Institute of Technology;Massachuse...
Poster
main
mixed reality;extended reality;robot manipulation;simulation;mujoco
https://openreview.net/forum?id=3p7rTnLJM8
-1
Lucid-XR: An Extended-Reality Data Engine for Robotic Manipulation We introduce Lucid-XR, a generative data engine for creating diverse and realistic-looking data to train real-world robot systems. At the core of Lucid-XR is vuer, a web-based physics simulation environment that runs directly on the XR headset, enabling...
[ -0.06853306293487549, -0.025825947523117065, -0.015237869694828987, 0.027021074667572975, -0.06080208718776703, 0.002675030380487442, 0.026498207822442055, 0.034621335566043854, 0.025396449491381645, 0.020597267895936966, -0.0032632569782435894, -0.026274120435118675, 0.0002710627450142056, ...
corl_2025_4IuTfpWGDR
4IuTfpWGDR
corl
2,025
TReF-6: Inferring Task-Relevant Frames from a Single Demonstration for One-Shot Skill Generalization
Robots often struggle to generalize from a single demonstration due to the lack of a transferable and interpretable spatial representation. In this work, we introduce TReF-6, a method that infers a simplified, abstracted 6DoF Task-Relevant Frame from a single trajectory. Our approach identifies an influence point purel...
Yuxuan Ding;Shuangge Wang;Tesca Fitzgerald
;Yale University;Yale University
Poster
main
Spatial Reference Frames;One-Shot Imitation Learning;Dynamic Movement Primitives
https://openreview.net/forum?id=4IuTfpWGDR
-1
TReF-6: Inferring Task-Relevant Frames from a Single Demonstration for One-Shot Skill Generalization Robots often struggle to generalize from a single demonstration due to the lack of a transferable and interpretable spatial representation. In this work, we introduce TReF-6, a method that infers a simplified, abstracte...
[ -0.013399561867117882, -0.01249714195728302, 0.005131940823048353, -0.030901029706001282, 0.02499428391456604, 0.012296603992581367, 0.010181843303143978, -0.002025885973125696, 0.014621018432080746, -0.00002702558231248986, -0.04298797994852066, -0.020035535097122192, -0.004963306710124016,...
corl_2025_4Po2mqLjrQ
4Po2mqLjrQ
corl
2,025
Motion Blender Gaussian Splatting for Dynamic Reconstruction
Gaussian splatting has emerged as a powerful tool for high-fidelity reconstruction of dynamic scenes. However, existing methods primarily rely on implicit motion representations, such as encoding motions into neural networks or per-Gaussian parameters, which makes it difficult to further manipulate the reconstructed m...
Xinyu Zhang;Haonan Chang;Yuhan Liu;Abdeslam Boularias
Rutgers University;;Rutgers University;, Rutgers University
Poster
main
Dynamic Reconstruction;Gaussian Splatting
https://openreview.net/forum?id=4Po2mqLjrQ
-1
Motion Blender Gaussian Splatting for Dynamic Reconstruction Gaussian splatting has emerged as a powerful tool for high-fidelity reconstruction of dynamic scenes. However, existing methods primarily rely on implicit motion representations, such as encoding motions into neural networks or per-Gaussian parameters, which...
[ -0.06705548614263535, -0.025478864088654518, -0.042224232107400894, -0.013100244104862213, -0.0020156889222562313, -0.03776497021317482, -0.00007697607361478731, -0.016671355813741684, 0.004355183336883783, 0.043408434838056564, -0.05695275589823723, 0.029790103435516357, 0.02914249151945114...
corl_2025_4XKKUifQ9c
4XKKUifQ9c
corl
2,025
ClutterDexGrasp: A Sim-to-Real System for General Dexterous Grasping in Cluttered Scenes
Dexterous grasping in cluttered scenes presents significant challenges due to diverse object geometries, occlusions, and potential collisions. Existing methods primarily focus on single-object grasping or grasp-pose prediction without interaction, which are insufficient for complex, cluttered scenes. Recent vision-lang...
Zeyuan Chen;Qiyang Yan;Yuanpei Chen;Tianhao Wu;Jiyao Zhang;Zihan Ding;Jinzhou Li;Yaodong Yang;Hao Dong
AgiBot+Peking University;AgiBot;PsiRobot;Peking University;Peking University;Princeton University;Duke University+Peking University;;Peking University+Peking University
Oral
main
Cluttered Scene;Dexterous Grasping;Sim-to-Real
https://openreview.net/forum?id=4XKKUifQ9c
-1
ClutterDexGrasp: A Sim-to-Real System for General Dexterous Grasping in Cluttered Scenes Dexterous grasping in cluttered scenes presents significant challenges due to diverse object geometries, occlusions, and potential collisions. Existing methods primarily focus on single-object grasping or grasp-pose prediction with...
[ -0.06330948323011398, -0.008872495032846928, -0.017707573249936104, -0.017557905986905098, -0.030326440930366516, -0.039212968200445175, -0.0061036404222249985, 0.01430263090878725, 0.0008453893242403865, -0.0036925857421010733, -0.0062813712283968925, 0.001993622165173292, 0.004415200557559...
corl_2025_4eMWCoWUKR
4eMWCoWUKR
corl
2,025
Enter the Mind Palace: Reasoning and Planning for Long-term Active Embodied Question Answering
As robots become increasingly capable of operating over extended periods—spanning days, weeks, and even months—they are expected to accumulate knowledge of their environments and leverage this experience to assist humans more effectively. This paper studies the problem of Long-term Active Embodied Question Answering (L...
Muhammad Fadhil Ginting;Dong-Ki Kim;Xiangyun Meng;Andrzej Marek Reinke;Bandi Jai Krishna;Navid Kayhani;Oriana Peltzer;David Fan;Amirreza Shaban;Sung-Kyun Kim;Mykel Kochenderfer;Ali-akbar Agha-mohammadi;Shayegan Omidshafiei
Stanford University;Field AI;University of Washington;;;Field AI;;Jet Propulsion Laboratory;University of Washington, Seattle;;Stanford University;;FieldAI
Poster
main
embodied question answering;long-term reasoning;vision-language navigation
https://openreview.net/forum?id=4eMWCoWUKR
-1
Enter the Mind Palace: Reasoning and Planning for Long-term Active Embodied Question Answering As robots become increasingly capable of operating over extended periods—spanning days, weeks, and even months—they are expected to accumulate knowledge of their environments and leverage this experience to assist humans more...
[ -0.05131544545292854, -0.02822534367442131, 0.015045511536300182, 0.016126127913594246, -0.03350835666060448, -0.05659846216440201, 0.0014558309921994805, 0.014629889279603958, -0.008612608537077904, 0.01234858762472868, 0.005033642519265413, 0.013826354406774044, -0.009166771546006203, 0....
corl_2025_4eSv0QeYlz
4eSv0QeYlz
corl
2,025
Deep Reactive Policy: Learning Reactive Manipulator Motion Planning for Dynamic Environments
Generating collision-free motion in dynamic, partially observable environments is a fundamental challenge for robotic manipulators. Classical motion planners can compute globally optimal trajectories but require full environment knowledge and are typically too slow for dynamic scenes. Neural motion policies offer a pro...
Jiahui Yang;Jason Jingzhou Liu;Yulong Li;Youssef Khaky;Kenneth Shaw;Deepak Pathak
Carnegie Mellon University;;Massachusetts Institute of Technology;Carnegie Mellon University;Carnegie Mellon University;Skild AI+Carnegie Mellon University
Poster
main
Motion Planning;Visuo-Motor Policy;Reactive Control
https://openreview.net/forum?id=4eSv0QeYlz
-1
Deep Reactive Policy: Learning Reactive Manipulator Motion Planning for Dynamic Environments Generating collision-free motion in dynamic, partially observable environments is a fundamental challenge for robotic manipulators. Classical motion planners can compute globally optimal trajectories but require full environmen...
[ -0.04432138800621033, -0.03052672930061817, 0.006888034753501415, 0.02507951110601425, -0.04130961745977402, -0.0004409597022458911, -0.019148921594023705, -0.003932035993784666, 0.04859736189246178, 0.026659762486815453, -0.01030881330370903, -0.01030881330370903, -0.019055966287851334, 0...
corl_2025_5htQM8jqOe
5htQM8jqOe
corl
2,025
D-Cubed: Latent Diffusion Trajectory Optimisation for Dexterous Deformable Manipulation
Mastering deformable object manipulation often necessitates the use of anthropomorphic, high-degree-of-freedom robot hands capable of precise, contact-rich control. However, current trajectory optimisation methods often struggle in these settings due to the large search space and the sparse task information available f...
Jun Yamada;Shaohong Zhong;Jack Collins;Ingmar Posner
University of Oxford;University of Oxford;University of Oxford;Amazon+University of Oxford
Poster
main
Trajectory Optimisation;Dexterous Deformable Object Manipulation;Latent Diffusion Model;Gradient-Free Guidance
https://openreview.net/forum?id=5htQM8jqOe
-1
D-Cubed: Latent Diffusion Trajectory Optimisation for Dexterous Deformable Manipulation Mastering deformable object manipulation often necessitates the use of anthropomorphic, high-degree-of-freedom robot hands capable of precise, contact-rich control. However, current trajectory optimisation methods often struggle in ...
[ -0.033578600734472275, -0.01627468690276146, -0.03797120600938797, 0.014207039028406143, -0.020069969817996025, 0.053262609988451004, -0.01946345902979374, -0.025473421439528465, 0.017110934481024742, 0.058702822774648666, -0.04940300062298775, -0.042124878615140915, 0.023561997339129448, ...
corl_2025_5ySSVlJBOn
5ySSVlJBOn
corl
2,025
FetchBot: Learning Generalizable Object Fetching in Cluttered Scenes via Zero-Shot Sim2Real
Generalizable object fetching in cluttered scenes remains a fundamental and application-critical challenge in embodied AI. Closely packed objects cause inevitable occlusions, making safe action generation particularly difficult. Under such partial observability, effective policies must not only generalize across divers...
Weiheng Liu;Yuxuan Wan;Jilong Wang;Yuxuan Kuang;Xuesong Shi;Haoran Li;Dongbin Zhao;Zhizheng Zhang;He Wang
Institute of Automation, Chinese Academy of Sciences;Peking University;Galbot Co. Ltd.;School of Computer Science, Carnegie Mellon University+Peking University;Galbot;Institute of Automation, Chinese Academy of Sciences;Institute of Automation, Chinese Academy of Sciences;Beijing Galbot Co., Ltd;Galbot+Peking Universit...
Oral
main
Generalizable Fetching;Sim2Real;Occlusion Handling
https://openreview.net/forum?id=5ySSVlJBOn
-1
FetchBot: Learning Generalizable Object Fetching in Cluttered Scenes via Zero-Shot Sim2Real Generalizable object fetching in cluttered scenes remains a fundamental and application-critical challenge in embodied AI. Closely packed objects cause inevitable occlusions, making safe action generation particularly difficult....
[ -0.027546469122171402, -0.038425859063863754, -0.020952899008989334, 0.01503700204193592, -0.005751057993620634, -0.03419498726725578, -0.02269286848604679, 0.003319679293781519, -0.020843006670475006, 0.021593941375613213, -0.04637477546930313, 0.01664876379072666, -0.020952899008989334, ...
corl_2025_6AASPlloSt
6AASPlloSt
corl
2,025
RICL: Adding In-Context Adaptability to Pre-Trained Vision-Language-Action Models
Multi-task ``vision-language-action'' (VLA) models have recently demonstrated increasing promise as generalist foundation models for robotics, achieving non-trivial performance out of the box on new tasks in new environments. However, for such models to be truly useful, an end user must have easy means to teach them to...
Kaustubh Sridhar;Souradeep Dutta;Dinesh Jayaraman;Insup Lee
Google Deepmind+University of Pennsylvania;University of British Columbia;University of Pennsylvania;University of Pennsylvania
Poster
main
Vision-Language-Action (VLA) models;In-Context Learning (ICL);Retrieval-Augmented Generation (RAG)
https://openreview.net/forum?id=6AASPlloSt
-1
RICL: Adding In-Context Adaptability to Pre-Trained Vision-Language-Action Models Multi-task ``vision-language-action'' (VLA) models have recently demonstrated increasing promise as generalist foundation models for robotics, achieving non-trivial performance out of the box on new tasks in new environments. However, for...
[ -0.04354650899767876, 0.011872134171426296, 0.013603678904473782, 0.013972092419862747, -0.030357295647263527, 0.02711525373160839, 0.02886522002518177, 0.017103610560297966, 0.004478530026972294, -0.015104965306818485, -0.03166516497731209, -0.037522945553064346, 0.008887982927262783, 0.0...
corl_2025_6yB6AX8aSU
6yB6AX8aSU
corl
2,025
LodeStar: Long-horizon Dexterity via Synthetic Data Augmentation from Human Demonstrations
Developing robotic systems capable of robustly executing long-horizon manipulation tasks with human-level dexterity is challenging, as such tasks require both physical dexterity and seamless sequencing of manipulation skills while robustly handling environment variations. While imitation learning offers a promising app...
Weikang Wan;Jiawei Fu;Xiaodi Yuan;Yifeng Zhu;Hao Su
University of California, San Diego;University of California, San Diego;University of California, San Diego;The University of Texas at Austin;University of California, San Diego
Poster
main
Dexterous Manipulation;Imitation Learning;Sim-to-Real
https://openreview.net/forum?id=6yB6AX8aSU
-1
LodeStar: Long-horizon Dexterity via Synthetic Data Augmentation from Human Demonstrations Developing robotic systems capable of robustly executing long-horizon manipulation tasks with human-level dexterity is challenging, as such tasks require both physical dexterity and seamless sequencing of manipulation skills whil...
[ -0.014004128985106945, -0.0175935011357069, -0.008359702304005623, -0.0027408478781580925, -0.021610621362924576, -0.014283095486462116, -0.0049842051230371, 0.01756560429930687, 0.004740109201520681, -0.010870402678847313, -0.03487083688378334, -0.0017830622382462025, -0.006002433598041534,...
corl_2025_7OOMC7pzaw
7OOMC7pzaw
corl
2,025
TypeTele: Releasing Dexterity in Teleoperation by Dexterous Manipulation Types
Dexterous teleoperation plays a crucial role in robotic manipulation for real-world data collection and remote robot control. Previous dexterous teleoperation mostly relies on hand retargeting to closely mimic human hand postures. However, these approaches may fail to fully leverage the inherent dexterity of dexterous ...
Yuhao Lin;Yi-Lin Wei;Haoran Liao;Mu Lin;Chengyi Xing;Hao Li;Dandan Zhang;Mark Cutkosky;Wei-Shi Zheng
SUN YAT-SEN UNIVERSITY;;;SUN YAT-SEN UNIVERSITY;Stanford University;Stanford University;Imperial College London;Stanford University;SUN YAT-SEN UNIVERSITY
Poster
main
Teleoperation;Dexterous;Manipulation
https://openreview.net/forum?id=7OOMC7pzaw
-1
TypeTele: Releasing Dexterity in Teleoperation by Dexterous Manipulation Types Dexterous teleoperation plays a crucial role in robotic manipulation for real-world data collection and remote robot control. Previous dexterous teleoperation mostly relies on hand retargeting to closely mimic human hand postures. However, t...
[ -0.013890476897358894, -0.013097262009978294, -0.02813144400715828, 0.03490143641829491, -0.01081907656043768, 0.050950195640325546, 0.025530437007546425, 0.025973161682486534, 0.021767280995845795, 0.004715475719422102, -0.028315911069512367, -0.021213876083493233, 0.00212945812381804, -0...
corl_2025_7XyO9Y1hI1
7XyO9Y1hI1
corl
2,025
EndoVLA: Dual-Phase Vision-Language-Action for Precise Autonomous Tracking in Endoscopy
In endoscopic procedures, autonomous tracking of abnormal regions and following of circumferential cutting markers can significantly reduce the cognitive burden on endoscopists. However, conventional model-based pipelines are fragile—each component (e.g., detection, motion planning) requires manual tuning and struggles...
CHI KIT NG;Long Bai;Guankun Wang;Yupeng Wang;Huxin Gao;Kun yuan;Chenhan Jin;Tieyong Zeng;Hongliang Ren
Chinese University of Hong Kong;;The Chinese University of Hong Kong;The Chinese University of Hong Kong;The Chinese University of Hong Kong;Université de Strasbourg+Technische Universität München;;The Chinese University of Hong Kong;The Chinese University of Hong Kong
Poster
main
Vision–Language–Action;Continuum Robots;Autonomous Endoscopic Tracking;Reinforcement Learning
https://openreview.net/forum?id=7XyO9Y1hI1
-1
EndoVLA: Dual-Phase Vision-Language-Action for Precise Autonomous Tracking in Endoscopy In endoscopic procedures, autonomous tracking of abnormal regions and following of circumferential cutting markers can significantly reduce the cognitive burden on endoscopists. However, conventional model-based pipelines are fragil...
[ -0.0641932561993599, -0.02579522505402565, 0.010235176421701908, 0.025961050763726234, -0.016923509538173676, 0.026919160038232803, -0.004873455036431551, 0.01867390051484108, 0.018627837300300598, 0.01839752309024334, 0.02408168464899063, -0.006679120473563671, 0.03439977392554283, 0.0265...
corl_2025_7iaYcss56y
7iaYcss56y
corl
2,025
ImMimic: Cross-Domain Imitation from Human Videos via Mapping and Interpolation
Learning robot manipulation from abundant human videos offers a scalable alternative to costly robot-specific data collection. However, domain gaps across visual, morphological, and physical aspects hinder direct imitation. To effectively bridge the domain gap, we propose ImMimic, an embodiment-agnostic co-training fra...
Yangcen Liu;Woo Chul Shin;Yunhai Han;Zhenyang Chen;Harish Ravichandar;Danfei Xu
Georgia Institute of Technology;Georgia Institute of Technology;Georgia Institute of Technology;;Georgia Institute of Technology;Georgia Institute of Technology+NVIDIA
Oral
main
Learning from Human;Imitation learning;Dexterous Manipulation
https://openreview.net/forum?id=7iaYcss56y
-1
ImMimic: Cross-Domain Imitation from Human Videos via Mapping and Interpolation Learning robot manipulation from abundant human videos offers a scalable alternative to costly robot-specific data collection. However, domain gaps across visual, morphological, and physical aspects hinder direct imitation. To effectively b...
[ -0.056169796735048294, -0.04429620876908302, -0.006143571808934212, 0.007756431587040424, -0.03030889853835106, -0.009180895052850246, 0.03644787520170212, -0.01736007072031498, -0.011129193007946014, 0.02446400560438633, -0.021651839837431908, -0.028599543496966362, 0.002678909804672003, ...
corl_2025_7wGYX11BJB
7wGYX11BJB
corl
2,025
ParticleFormer: A 3D Point Cloud World Model for Multi-Object, Multi-Material Robotic Manipulation
3D world models (i.e., learning-based 3D dynamics models) offer a promising approach to generalizable robotic manipulation by capturing the underlying physics of environment evolution conditioned on robot actions. However, existing 3D world models are primarily limited to single-material dynamics using a particle-based...
Suning Huang;Qianzhong Chen;Xiaohan Zhang;Jiankai Sun;Mac Schwager
;Stanford University;Boston Dynamics AI Institute;Stanford University;Stanford University
Poster
main
Learning-based Dynamics Modeling;Model-based Planning
https://openreview.net/forum?id=7wGYX11BJB
-1
ParticleFormer: A 3D Point Cloud World Model for Multi-Object, Multi-Material Robotic Manipulation 3D world models (i.e., learning-based 3D dynamics models) offer a promising approach to generalizable robotic manipulation by capturing the underlying physics of environment evolution conditioned on robot actions. However...
[ -0.017406122758984566, -0.013297226279973984, -0.03896649554371834, -0.028227126225829124, -0.021242907270789146, -0.0041678547859191895, -0.007487626280635595, 0.011301735416054726, -0.0025306453462690115, 0.00455108005553484, -0.020499134436249733, -0.016635138541460037, 0.0298960823565721...
corl_2025_8DHSyMFLbB
8DHSyMFLbB
corl
2,025
Sim-to-Real Reinforcement Learning for Vision-Based Dexterous Manipulation on Humanoids
Learning generalizable robot manipulation policies, especially for complex multi-fingered humanoids, remains a significant challenge. Existing approaches primarily rely on extensive data collection and imitation learning, which are expensive, labor-intensive, and difficult to scale. Sim-to-real reinforcement learning (...
Toru Lin;Kartik Sachdev;Linxi Fan;Jitendra Malik;Yuke Zhu
;NVIDIA;NVIDIA;Meta Facebook+University of California, Berkeley;Computer Science Department, University of Texas, Austin
Poster
main
Humanoids;Vision-Based Dexterous Manipulation;Reinforcement Learning;Sim-to-Real
https://openreview.net/forum?id=8DHSyMFLbB
-1
Sim-to-Real Reinforcement Learning for Vision-Based Dexterous Manipulation on Humanoids Learning generalizable robot manipulation policies, especially for complex multi-fingered humanoids, remains a significant challenge. Existing approaches primarily rely on extensive data collection and imitation learning, which are ...
[ -0.037713438272476196, -0.006729149725288153, 0.017950691282749176, -0.020574398338794708, -0.04129980131983757, -0.03369293734431267, 0.01584606245160103, 0.029087292030453682, 0.019800499081611633, -0.004851027857512236, -0.02342461235821247, 0.0031262703705579042, 0.016459519043564796, ...
corl_2025_8RdxHk9hpr
8RdxHk9hpr
corl
2,025
Dynamics-Compliant Trajectory Diffusion for Super-Nominal Payload Manipulation
Nominal payload ratings for articulated robots are typically derived from worst-case configurations, resulting in uniform payload constraints across the entire workspace. This conservative approach severely underutilizes the robot's inherent capabilities---our analysis demonstrates that manipulators can safely handle p...
Anuj Pasricha;Joewie J. Koh;Jay Vakil;Alessandro Roncone
University of Colorado at Boulder;;University of Colorado at Boulder;University of Colorado at Boulder
Poster
main
Robot Planning;Grasping & Manipulation;Robot Modeling & Simulation;diffusion models;dynamics-constrained planning;payload transport
https://openreview.net/forum?id=8RdxHk9hpr
-1
Dynamics-Compliant Trajectory Diffusion for Super-Nominal Payload Manipulation Nominal payload ratings for articulated robots are typically derived from worst-case configurations, resulting in uniform payload constraints across the entire workspace. This conservative approach severely underutilizes the robot's inherent...
[ -0.08853457868099213, -0.04746070131659508, 0.0024248820263892412, 0.0332004651427269, -0.03349411487579346, -0.01443458627909422, 0.019857879728078842, -0.004542352631688118, 0.016123056411743164, 0.04033976048231125, -0.02661176212131977, 0.011562352068722248, -0.0038311760872602463, 0.0...
corl_2025_8v0mlyKk5q
8v0mlyKk5q
corl
2,025
Beyond Constant Parameters: Hyper Prediction Models and HyperMPC
Model Predictive Control (MPC) is among the most widely adopted and reliable methods for robot control, relying critically on an accurate dynamics model. However, existing dynamics models used in the gradient-based MPC are limited by computational complexity and state representation. To address this limitation, we prop...
Jan Węgrzynowski;Piotr Kicki;Grzegorz Czechmanowski;Maciej Piotr Krupka;Krzysztof Walas
Technical University of Poznan;Technical University of Poznan+IDEAS NCBR Sp.;;;Technical University of Poznan
Poster
main
Model Learning for Robot Control;Dynamics Model Learning;Model Predictive Control;MPC
https://openreview.net/forum?id=8v0mlyKk5q
-1
Beyond Constant Parameters: Hyper Prediction Models and HyperMPC Model Predictive Control (MPC) is among the most widely adopted and reliable methods for robot control, relying critically on an accurate dynamics model. However, existing dynamics models used in the gradient-based MPC are limited by computational complex...
[ -0.052498433738946915, -0.04593612998723984, -0.02565264329314232, 0.04090254381299019, -0.018586980178952217, -0.051566287875175476, 0.014410968869924545, 0.020134342834353447, 0.029679512605071068, 0.02841179445385933, -0.07807650417089462, -0.03258780762553215, 0.018978482112288475, 0.0...
corl_2025_93bWCbhXJR
93bWCbhXJR
corl
2,025
Distilling On-device Language Models for Robot Planning with Minimal Human Intervention
Large language models (LLMs) provide robots with powerful contextual reasoning abilities and a natural human interface. Yet, current LLM-enabled robots typically depend on cloud-hosted models, limiting their usability in environments with unreliable communication infrastructure, such as outdoor or industrial settings. ...
Zachary Ravichandran;Ignacio Hounie;Fernando Cladera;Alejandro Ribeiro;George J. Pappas;Vijay Kumar
University of Pennsylvania;University of Pennsylvania;University of Pennsylvania;University of Pennsylvania;;
Poster
main
LLM-enabled Robots;LLM Distillation
https://openreview.net/forum?id=93bWCbhXJR
-1
Distilling On-device Language Models for Robot Planning with Minimal Human Intervention Large language models (LLMs) provide robots with powerful contextual reasoning abilities and a natural human interface. Yet, current LLM-enabled robots typically depend on cloud-hosted models, limiting their usability in environment...
[ -0.0357489250600338, -0.00746771739795804, -0.01354910060763359, -0.032514143735170364, -0.026876388117671013, 0.002130332635715604, 0.009778273291885853, -0.01304077822715044, 0.004727397579699755, 0.028706347569823265, -0.03985247015953064, 0.032828379422426224, -0.0031908778473734856, 0...
corl_2025_9AHjtHLlIe
9AHjtHLlIe
corl
2,025
Imagine, Verify, Execute: Memory-guided Agentic Exploration with Vision-Language Models
Exploration is key for general-purpose robotic learning, particularly in open-ended environments where explicit guidance or task-specific feedback is limited. Vision-language models (VLMs), which can reason about object semantics, spatial relations, and potential outcomes, offer a promising foundation for guiding explo...
Seungjae Lee;Daniel Ekpo;Haowen Liu;Furong Huang;Abhinav Shrivastava;Jia-Bin Huang
University of Maryland, College Park;University of Maryland, College Park;;University of Maryland;Department of Computer Science, University of Maryland, College Park;University of Maryland, College Park
Poster
main
Exploration;Agentic System;Vision-Language Model
https://openreview.net/forum?id=9AHjtHLlIe
-1
Imagine, Verify, Execute: Memory-guided Agentic Exploration with Vision-Language Models Exploration is key for general-purpose robotic learning, particularly in open-ended environments where explicit guidance or task-specific feedback is limited. Vision-language models (VLMs), which can reason about object semantics, s...
[ -0.03326142206788063, -0.041755300015211105, 0.02925877831876278, -0.005111359991133213, -0.014037447981536388, -0.02557559125125408, 0.00871937908232212, 0.0014234761474654078, 0.017166277393698692, -0.024091042578220367, -0.0010723051382228732, -0.0631779134273529, -0.01999443769454956, ...
corl_2025_9FpccnRarn
9FpccnRarn
corl
2,025
BEVCalib: LiDAR-Camera Calibration via Geometry-Guided Bird’s-Eye View Representation
Accurate LiDAR-camera calibration is the foundation of accurate multimodal fusion environmental perception for autonomous driving and robotic systems. Traditional calibration methods require extensive data collection in controlled environments and cannot compensate for the transformation changes during the vehicle/robo...
Weiduo Yuan;Jerry Li;Justin Yue;Divyank Shah;Konstantinos Karydis;Hang Qiu
University of Southern California;University of Southern California+University of California, Riverside;University of California, Riverside;University of California, Riverside;;University of California, Riverside
Poster
main
LiDAR-Camera Calibration;Autonomous Driving;BEV Features
https://openreview.net/forum?id=9FpccnRarn
-1
BEVCalib: LiDAR-Camera Calibration via Geometry-Guided Bird’s-Eye View Representation Accurate LiDAR-camera calibration is the foundation of accurate multimodal fusion environmental perception for autonomous driving and robotic systems. Traditional calibration methods require extensive data collection in controlled env...
[ -0.05215809866786003, -0.034574639052152634, -0.02955872379243374, 0.023376749828457832, -0.003058596746996045, 0.031631723046302795, 0.010670384392142296, -0.008773221634328365, 0.02768932469189167, 0.025875451043248177, -0.024302193894982338, -0.05249125882983208, -0.009180417284369469, ...