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_2023_-3G6_D66Aua
-3G6_D66Aua
corl
2,023
Simultaneous Learning of Contact and Continuous Dynamics
Robotic manipulation can greatly benefit from the data efficiency, robustness, and predictability of model-based methods if robots can quickly generate models of novel objects they encounter. This is especially difficult when effects like complex joint friction lack clear first-principles models and are usually ignored...
Bibit Bianchini;Mathew Halm;Michael Posa
University of Pennsylvania;School of Engineering and Applied Science, University of Pennsylvania;University of Pennsylvania
Poster
main
system identification;dynamics learning;contact-rich manipulation
https://github.com/ebianchi/dair_pll
https://openreview.net/forum?id=-3G6_D66Aua
14
Simultaneous Learning of Contact and Continuous Dynamics Robotic manipulation can greatly benefit from the data efficiency, robustness, and predictability of model-based methods if robots can quickly generate models of novel objects they encounter. This is especially difficult when effects like complex joint friction l...
[ -0.04371373727917671, -0.018341166898608208, -0.028888268396258354, 0.036552123725414276, -0.01290950272232294, -0.016406601294875145, 0.014174411073327065, 0.010109963826835155, -0.0027111817616969347, 0.028813861310482025, -0.0283674243837595, -0.039956215769052505, -0.0002644913620315492,...
corl_2023_-HFJuX1uqs
-HFJuX1uqs
corl
2,023
Act3D: 3D Feature Field Transformers for Multi-Task Robotic Manipulation
3D perceptual representations are well suited for robot manipulation as they easily encode occlusions and simplify spatial reasoning. Many manipulation tasks require high spatial precision in end-effector pose prediction, which typically demands high-resolution 3D feature grids that are computationally expensive to pro...
Theophile Gervet;Zhou Xian;Nikolaos Gkanatsios;Katerina Fragkiadaki
Carnegie Mellon University;Carnegie Mellon University;Carnegie Mellon University;Carnegie Mellon University
Poster
main
Learning from Demonstrations;Manipulation;Transformers
https://github.com/zhouxian/chained-diffuser
https://openreview.net/forum?id=-HFJuX1uqs
71
Act3D: 3D Feature Field Transformers for Multi-Task Robotic Manipulation 3D perceptual representations are well suited for robot manipulation as they easily encode occlusions and simplify spatial reasoning. Many manipulation tasks require high spatial precision in end-effector pose prediction, which typically demands h...
[ -0.034714847803115845, -0.04591735452413559, -0.02775711379945278, 0.006344087887555361, -0.0011136757675558329, -0.020153436809778214, -0.0030197857413440943, -0.01837247796356678, 0.03823985531926155, 0.007105378434062004, -0.01246670912951231, 0.02094702422618866, 0.016083993017673492, ...
corl_2023_-K7-1WvKO3F
-K7-1WvKO3F
corl
2,023
ViNT: A Foundation Model for Visual Navigation
General-purpose pre-trained models (``foundation models'') have enabled practitioners to produce generalizable solutions for individual machine learning problems with datasets that are significantly smaller than those required for learning from scratch. Such models are typically trained on large and diverse datasets wi...
Dhruv Shah;Ajay Sridhar;Nitish Dashora;Kyle Stachowicz;Kevin Black;Noriaki Hirose;Sergey Levine
UC Berkeley;University of California, Berkeley;University of California, Berkeley;University of California, Berkeley;University of California, Berkeley;Toyota Central R&D Labs., Inc;Google
Oral
main
visual navigation;multi-task learning;planning;generalization
https://github.com/robodhruv/visualnav-transformer
https://openreview.net/forum?id=-K7-1WvKO3F
154
ViNT: A Foundation Model for Visual Navigation General-purpose pre-trained models (``foundation models'') have enabled practitioners to produce generalizable solutions for individual machine learning problems with datasets that are significantly smaller than those required for learning from scratch. Such models are typ...
[ -0.047531452029943466, -0.057111434638500214, -0.03872523456811905, 0.0334378220140934, -0.0211496539413929, -0.010409018956124783, 0.013614628463983536, -0.004048463888466358, 0.000013853265045327134, 0.02474214695394039, -0.014637107960879803, -0.02997429110109806, 0.00010247817408526316, ...
corl_2023_09UL1dCqf2n
09UL1dCqf2n
corl
2,023
Preference learning for guiding the tree search in continuous POMDPs
A robot operating in a partially observable environment must perform sensing actions to achieve a goal, such as clearing the objects in front of a shelf to better localize a target object at the back, and estimate its shape for grasping. A POMDP is a principled framework for enabling robots to perform such information-...
Jiyong Ahn;Sanghyeon Son;Dongryung Lee;Jisu Han;Dongwon Son;Beomjoon Kim
Korea Advanced Institute of Science & Technology;;Korea Advanced Institute of Science & Technology;Korea Advanced Institute of Science & Technology;KAIST;Korea Advanced Institute of Science & Technology
Poster
main
POMDP;Online planning;Guided Search;Preference-based learning
https://openreview.net/forum?id=09UL1dCqf2n
0
Preference learning for guiding the tree search in continuous POMDPs A robot operating in a partially observable environment must perform sensing actions to achieve a goal, such as clearing the objects in front of a shelf to better localize a target object at the back, and estimate its shape for grasping. A POMDP is a ...
[ -0.06318318843841553, -0.036296725273132324, -0.02516871690750122, -0.0005213336553424597, -0.006226830184459686, -0.047387391328811646, 0.02313356101512909, -0.006539572030305862, 0.04074046015739441, 0.02350698411464691, -0.018521785736083984, -0.046864598989486694, -0.009774349629878998, ...
corl_2023_0I3su3mkuL
0I3su3mkuL
corl
2,023
Q-Transformer: Scalable Offline Reinforcement Learning via Autoregressive Q-Functions
In this work, we present a scalable reinforcement learning method for training multi-task policies from large offline datasets that can leverage both human demonstrations and autonomously collected data. Our method uses a Transformer to provide a scalable representation for Q-functions trained via offline temporal diff...
Yevgen Chebotar;Quan Vuong;Karol Hausman;Fei Xia;Yao Lu;Alex Irpan;Aviral Kumar;Tianhe Yu;Alexander Herzog;Karl Pertsch;Keerthana Gopalakrishnan;Julian Ibarz;Ofir Nachum;Sumedh Anand Sontakke;Grecia Salazar;Huong T Tran;Jodilyn Peralta;Clayton Tan;Deeksha Manjunath;Jaspiar Singh;Brianna Zitkovich;Tomas Jackson;Kanishka...
Google;;;Google;Google;Google DeepMind;University of California, Berkeley;Google Brain;Google;University of Southern California;Research, Google;Google;OpenAI;University of Southern California;;;;;;;;;;Google;Google
Poster
main
Reinforcement Learning;Offline RL;Transformers;Q-Learning;Robotic Manipulation
https://openreview.net/forum?id=0I3su3mkuL
106
Q-Transformer: Scalable Offline Reinforcement Learning via Autoregressive Q-Functions In this work, we present a scalable reinforcement learning method for training multi-task policies from large offline datasets that can leverage both human demonstrations and autonomously collected data. Our method uses a Transformer ...
[ -0.052003927528858185, 0.010328659787774086, -0.01656099408864975, 0.0031161669176071882, -0.02724103257060051, 0.012779058888554573, 0.04863809421658516, 0.02093472331762314, 0.0022677744273096323, 0.015238704159855843, -0.0039506894536316395, 0.009316137060523033, -0.00492391362786293, 0...
corl_2023_0bZaUfELuW
0bZaUfELuW
corl
2,023
Goal Representations for Instruction Following: A Semi-Supervised Language Interface to Control
Our goal is for robots to follow natural language instructions like ``put the towel next to the microwave.'' But getting large amounts of labeled data, i.e. data that contains demonstrations of tasks labeled with the language instruction, is prohibitive. In contrast, obtaining policies that respond to image goals is mu...
Vivek Myers;Andre Wang He;Kuan Fang;Homer Rich Walke;Philippe Hansen-Estruch;Ching-An Cheng;Mihai Jalobeanu;Andrey Kolobov;Anca Dragan;Sergey Levine
University of California, Berkeley;UC Berkeley, University of California, Berkeley;;University of California, Berkeley;;Microsoft Research;Microsoft Research;Microsoft;University of California, Berkeley;Google
Poster
main
Instruction Following;Representation Learning;Manipulation
https://github.com/rail-berkeley/grif_release
https://openreview.net/forum?id=0bZaUfELuW
32
Goal Representations for Instruction Following: A Semi-Supervised Language Interface to Control Our goal is for robots to follow natural language instructions like ``put the towel next to the microwave.'' But getting large amounts of labeled data, i.e. data that contains demonstrations of tasks labeled with the languag...
[ -0.05153782665729523, -0.009409105405211449, -0.010113505646586418, 0.009824281558394432, -0.040715258568525314, -0.03181462734937668, -0.01010417565703392, -0.0037855671253055334, 0.01114911399781704, 0.023455122485756874, 0.011634263209998608, -0.027112405747175217, 0.029388876631855965, ...
corl_2023_0hPkttoGAf
0hPkttoGAf
corl
2,023
RVT: Robotic View Transformer for 3D Object Manipulation
For 3D object manipulation, methods that build an explicit 3D representation perform better than those relying only on camera images. But using explicit 3D representations like voxels comes at large computing cost, adversely affecting scalability. In this work, we propose RVT, a multi-view transformer for 3D manipulati...
Ankit Goyal;Jie Xu;Yijie Guo;Valts Blukis;Yu-Wei Chao;Dieter Fox
NVIDIA;NVIDIA;University of Michigan;NVIDIA;NVIDIA;Department of Computer Science
Oral
main
3D Manipulation;Multi-View;Transformer
https://github.com/nvlabs/rvt
https://openreview.net/forum?id=0hPkttoGAf
140
RVT: Robotic View Transformer for 3D Object Manipulation For 3D object manipulation, methods that build an explicit 3D representation perform better than those relying only on camera images. But using explicit 3D representations like voxels comes at large computing cost, adversely affecting scalability. In this work, w...
[ -0.035961247980594635, -0.002318314043805003, -0.006749579682946205, 0.03733033314347267, -0.01696750335395336, 0.057428471744060516, -0.012476909905672073, 0.010249868966639042, 0.021832311525940895, -0.026286395266652107, -0.023840298876166344, -0.009903034195303917, 0.035395361483097076, ...
corl_2023_0hQMcWfjG9
0hQMcWfjG9
corl
2,023
$\alpha$-MDF: An Attention-based Multimodal Differentiable Filter for Robot State Estimation
Differentiable Filters are recursive Bayesian estimators that derive the state transition and measurement models from data alone. Their data-driven nature eschews the need for explicit analytical models, while remaining algorithmic components of the filtering process intact. As a result, the gain mechanism -- a critica...
Xiao Liu;Yifan Zhou;Shuhei Ikemoto;Heni Ben Amor
Arizona State University;Arizona State University;Kyushu Institute of Technology;Arizona State University
Poster
main
Differentiable Filters;Sensor Fusion;Multimodal Learning
https://github.com/ir-lab/alpha-MDF
https://openreview.net/forum?id=0hQMcWfjG9
8
$\alpha$-MDF: An Attention-based Multimodal Differentiable Filter for Robot State Estimation Differentiable Filters are recursive Bayesian estimators that derive the state transition and measurement models from data alone. Their data-driven nature eschews the need for explicit analytical models, while remaining algorit...
[ -0.008578325621783733, -0.037264786660671234, -0.05251111835241318, -0.02297814190387726, -0.009868469089269638, -0.00478937616571784, 0.02991323173046112, 0.017781352624297142, 0.04678921774029732, 0.0035852419678121805, -0.048672378063201904, -0.012584561482071877, -0.04573899507522583, ...
corl_2023_0mRSANSzEK
0mRSANSzEK
corl
2,023
Improving Behavioural Cloning with Positive Unlabeled Learning
Learning control policies offline from pre-recorded datasets is a promising avenue for solving challenging real-world problems. However, available datasets are typically of mixed quality, with a limited number of the trajectories that we would consider as positive examples; i.e., high-quality demonstrations. Therefore,...
Qiang Wang;Robert McCarthy;David Cordova Bulens;Kevin McGuinness;Noel E. O’Connor;Francisco Roldan Sanchez;Nico Gürtler;Felix Widmaier;Stephen J. Redmond
University College Dublin;;;Dublin City University;;Insight Centre for Data Analytics;Max Planck Institute for Intelligent Systems, Max-Planck Institute;, Max Planck Institute for Intelligent Systems;
Poster
main
Offline policy learning;positive unlabeled learning;behavioural cloning
https://openreview.net/forum?id=0mRSANSzEK
8
Improving Behavioural Cloning with Positive Unlabeled Learning Learning control policies offline from pre-recorded datasets is a promising avenue for solving challenging real-world problems. However, available datasets are typically of mixed quality, with a limited number of the trajectories that we would consider as p...
[ -0.08705994486808777, -0.03501727804541588, -0.01797330565750599, 0.014562653377652168, -0.03202482685446739, 0.013521799817681313, 0.024069733917713165, -0.007095459848642349, -0.032656773924827576, 0.023698000237345695, -0.027972932904958725, -0.025426559150218964, -0.005799040198326111, ...
corl_2023_0o2JgvlzMUc
0o2JgvlzMUc
corl
2,023
Deception Game: Closing the Safety-Learning Loop in Interactive Robot Autonomy
An outstanding challenge for the widespread deployment of robotic systems like autonomous vehicles is ensuring safe interaction with humans without sacrificing performance. Existing safety methods often neglect the robot’s ability to learn and adapt at runtime, leading to overly conservative behavior. This paper propos...
Haimin Hu;Zixu Zhang;Kensuke Nakamura;Andrea Bajcsy;Jaime Fernández Fisac
Toyota Research Institute;Princeton University;Princeton University;Princeton University;University of California, Berkeley
Poster
main
Learning-Aware Safety Analysis;Active Information Gathering;Adversarial Reinforcement Learning
https://openreview.net/forum?id=0o2JgvlzMUc
16
Deception Game: Closing the Safety-Learning Loop in Interactive Robot Autonomy An outstanding challenge for the widespread deployment of robotic systems like autonomous vehicles is ensuring safe interaction with humans without sacrificing performance. Existing safety methods often neglect the robot’s ability to learn a...
[ -0.039126694202423096, -0.019563347101211548, -0.008261033333837986, 0.019830292090773582, -0.04301648214459419, -0.03264371305704117, 0.021012483164668083, 0.007770042400807142, -0.007960718125104904, 0.014577171765267849, -0.023167120292782784, -0.019067589193582535, 0.003251024056226015, ...
corl_2023_2Qrd-Yw4YmF
2Qrd-Yw4YmF
corl
2,023
Sequential Dexterity: Chaining Dexterous Policies for Long-Horizon Manipulation
Many real-world manipulation tasks consist of a series of subtasks that are significantly different from one another. Such long-horizon, complex tasks highlight the potential of dexterous hands, which possess adaptability and versatility, capable of seamlessly transitioning between different modes of functionality with...
Yuanpei Chen;Chen Wang;Li Fei-Fei;Karen Liu
South China University of Technology;Computer Science Department, Stanford University;Stanford University;Computer Science Department, Stanford University
Poster
main
Dexterous Manipulation;Reinforcement Learning;Long-Horizon Manipulation
https://github.com/sequential-dexterity/SeqDex
https://openreview.net/forum?id=2Qrd-Yw4YmF
48
Sequential Dexterity: Chaining Dexterous Policies for Long-Horizon Manipulation Many real-world manipulation tasks consist of a series of subtasks that are significantly different from one another. Such long-horizon, complex tasks highlight the potential of dexterous hands, which possess adaptability and versatility, c...
[ -0.045530855655670166, -0.006325271911919117, -0.011179120279848576, 0.003926110919564962, -0.0346478708088398, -0.000934099720325321, 0.006362289190292358, 0.029446987435221672, -0.003451830940321088, -0.007259950041770935, -0.032093700021505356, -0.0005445540882647038, 0.015482340939342976...
corl_2023_2qKBwyLnln
2qKBwyLnln
corl
2,023
Policy Stitching: Learning Transferable Robot Policies
Training robots with reinforcement learning (RL) typically involves heavy interactions with the environment, and the acquired skills are often sensitive to changes in task environments and robot kinematics. Transfer RL aims to leverage previous knowledge to accelerate learning of new tasks or new body configurations. H...
Pingcheng Jian;Easop Lee;Zachary Bell;Michael M. Zavlanos;Boyuan Chen
Duke University;Duke University;;;Duke University
Poster
main
robot transfer learning;policy stitching
https://github.com/general-robotics-duke/Policy-Stitching
https://openreview.net/forum?id=2qKBwyLnln
9
Policy Stitching: Learning Transferable Robot Policies Training robots with reinforcement learning (RL) typically involves heavy interactions with the environment, and the acquired skills are often sensitive to changes in task environments and robot kinematics. Transfer RL aims to leverage previous knowledge to acceler...
[ -0.07987476885318756, -0.030967291444540024, -0.02274160459637642, 0.03132088482379913, -0.01488812081515789, -0.010468210093677044, -0.005224800202995539, 0.01800532080233097, 0.030297325924038887, 0.023616282269358635, -0.036996982991695404, -0.007858136668801308, 0.02661251649260521, 0....
corl_2023_32c8pl84_uD
32c8pl84_uD
corl
2,023
Marginalized Importance Sampling for Off-Environment Policy Evaluation
Reinforcement Learning (RL) methods are typically sample-inefficient, making it challenging to train and deploy RL-policies in real world robots. Even a robust policy trained in simulation requires a real-world deployment to assess their performance. This paper proposes a new approach to evaluate the real-world perform...
Pulkit Katdare;Nan Jiang;Katherine Rose Driggs-Campbell
University of Illinois, Urbana Champaign;University of Illinois, Urbana Champaign;
Poster
main
Sim2Real;Policy Evaluation;Robot Validation
https://github.com/pulkitkatdare/mis_off_env_eval
https://openreview.net/forum?id=32c8pl84_uD
5
Marginalized Importance Sampling for Off-Environment Policy Evaluation Reinforcement Learning (RL) methods are typically sample-inefficient, making it challenging to train and deploy RL-policies in real world robots. Even a robust policy trained in simulation requires a real-world deployment to assess their performance...
[ -0.04497583955526352, 0.006974159274250269, -0.003999818582087755, 0.02272387035191059, -0.02150781638920307, -0.037171315401792526, 0.04109172895550728, 0.005939606111496687, -0.000502529670484364, 0.070131815969944, -0.06443270295858383, -0.02272387035191059, -0.006747283507138491, 0.027...
corl_2023_3gh9hf3R6x
3gh9hf3R6x
corl
2,023
Robot Learning with Sensorimotor Pre-training
We present a self-supervised sensorimotor pre-training approach for robotics. Our model, called RPT, is a Transformer that operates on sequences of sensorimotor tokens. Given a sequence of camera images, proprioceptive robot states, and actions, we encode the sequence into tokens, mask out a subset, and train a model t...
Ilija Radosavovic;Baifeng Shi;Letian Fu;Ken Goldberg;Trevor Darrell;Jitendra Malik
University of California, Berkeley;University of California, Berkeley;Electrical Engineering & Computer Science Department, University of California, Berkeley;University of California, Berkeley;University of California, Berkeley;Electrical Engineering & Computer Science Department
Oral
main
Robot Learning;Self-supervised;Sensorimotor;Pre-training
https://github.com/ir413/rpt
https://openreview.net/forum?id=3gh9hf3R6x
54
Robot Learning with Sensorimotor Pre-training We present a self-supervised sensorimotor pre-training approach for robotics. Our model, called RPT, is a Transformer that operates on sequences of sensorimotor tokens. Given a sequence of camera images, proprioceptive robot states, and actions, we encode the sequence into ...
[ -0.053135816007852554, -0.01490145456045866, 0.0006245736149139702, 0.029393885284662247, -0.00836638081818819, -0.013572129420936108, -0.011731525883078575, 0.01898238994181156, -0.0027864696457982063, 0.018517591059207916, -0.03203394263982773, 0.001796447904780507, -0.011136583052575588, ...
corl_2023_3uwj8QZROL
3uwj8QZROL
corl
2,023
Scaling Up and Distilling Down: Language-Guided Robot Skill Acquisition
We present a framework for robot skill acquisition, which 1) efficiently scale up data generation of language-labelled robot data and 2) effectively distills this data down into a robust multi-task language-conditioned visuo-motor policy. For (1), we use a large language model (LLM) to guide high-level planning, and sa...
Huy Ha;Pete Florence;Shuran Song
Columbia University;Google;Columbia University
Poster
main
skill learning;language;diffusion
https://github.com/real-stanford/scalingup
https://openreview.net/forum?id=3uwj8QZROL
170
Scaling Up and Distilling Down: Language-Guided Robot Skill Acquisition We present a framework for robot skill acquisition, which 1) efficiently scale up data generation of language-labelled robot data and 2) effectively distills this data down into a robust multi-task language-conditioned visuo-motor policy. For (1), ...
[ -0.06160910427570343, -0.031263772398233414, -0.0143552515655756, 0.01069985143840313, -0.02672666870057583, -0.00417432002723217, -0.05051432177424431, 0.007173033896833658, -0.016311533749103546, 0.0373622290790081, -0.029445257037878036, -0.010369212366640568, -0.016045184805989265, 0.0...
corl_2023_44FPaVRWkbl
44FPaVRWkbl
corl
2,023
DORT: Modeling Dynamic Objects in Recurrent for Multi-Camera 3D Object Detection and Tracking
Recent multi-camera 3D object detectors usually leverage temporal information to construct multi-view stereo that alleviates the ill-posed depth estimation. However, they typically assume all the objects are static and directly aggregate features across frames. This work begins with a theoretical and empirical analysis...
Qing LIAN;Tai Wang;Dahua Lin;Jiangmiao Pang
Hong Kong University of Science and Technology;The Chinese University of Hong Kong;Shanghai AI Laboratory ;The Chinese University of Hong Kong
Poster
main
Temporal Modeling;3D Object Detection
https://github.com/OpenRobotLab/DORT
https://openreview.net/forum?id=44FPaVRWkbl
10
DORT: Modeling Dynamic Objects in Recurrent for Multi-Camera 3D Object Detection and Tracking Recent multi-camera 3D object detectors usually leverage temporal information to construct multi-view stereo that alleviates the ill-posed depth estimation. However, they typically assume all the objects are static and directl...
[ -0.03586338832974434, -0.011432592757046223, -0.036191076040267944, 0.006508203689008951, 0.0027238880284130573, 0.02299262024462223, -0.023029029369354248, -0.0444924496114254, 0.01571071334183216, -0.0033587792422622442, -0.04157968610525131, -0.008305924944579601, -0.005839178804308176, ...
corl_2023_48qUHKUEdBf
48qUHKUEdBf
corl
2,023
STOW: Discrete-Frame Segmentation and Tracking of Unseen Objects for Warehouse Picking Robots
Segmentation and tracking of unseen object instances in discrete frames pose a significant challenge in dynamic industrial robotic contexts, such as distribution warehouses. Here, robots must handle object rearrangements, including shifting, removal, and partial occlusion by new items, and track these items after subst...
Yi Li;Muru Zhang;Markus Grotz;Kaichun Mo;Dieter Fox
Department of Computer Science, University of Washington;University of Washington;University of Washington;NVIDIA;Department of Computer Science
Poster
main
Unseen Object Instance Segmentation;Unsupervised Multi Object Tracking;Zero-shot;Discrete Frames
https://sites.google.com/view/stow-corl23
https://openreview.net/forum?id=48qUHKUEdBf
6
STOW: Discrete-Frame Segmentation and Tracking of Unseen Objects for Warehouse Picking Robots Segmentation and tracking of unseen object instances in discrete frames pose a significant challenge in dynamic industrial robotic contexts, such as distribution warehouses. Here, robots must handle object rearrangements, incl...
[ -0.005947257857769728, -0.030816342681646347, -0.01780005358159542, 0.037250321358442307, -0.009173517115414143, 0.02805362641811371, 0.010253572836518288, 0.008598724380135536, 0.0217865239828825, 0.04717014357447624, -0.0032726142089813948, -0.010327739641070366, 0.002328145084902644, 0....
corl_2023_4ZK8ODNyFXx
4ZK8ODNyFXx
corl
2,023
Robots That Ask For Help: Uncertainty Alignment for Large Language Model Planners
Large language models (LLMs) exhibit a wide range of promising capabilities --- from step-by-step planning to commonsense reasoning --- that may provide utility for robots, but remain prone to confidently hallucinated predictions. In this work, we present KnowNo, a framework for measuring and aligning the uncertainty o...
Allen Z. Ren;Anushri Dixit;Alexandra Bodrova;Sumeet Singh;Stephen Tu;Noah Brown;Peng Xu;Leila Takayama;Fei Xia;Jake Varley;Zhenjia Xu;Dorsa Sadigh;Andy Zeng;Anirudha Majumdar
Google DeepMind;California Institute of Technology;Princeton University;Google Brain Robotics;Google;Research, Google;Google;;Google;Google;Columbia University;Stanford University;Princeton University;Google
Oral
main
Language-based planning;uncertainty estimation;conformal prediction
https://github.com/google-research/google-research/tree/master/language_model_uncertainty
https://openreview.net/forum?id=4ZK8ODNyFXx
248
Robots That Ask For Help: Uncertainty Alignment for Large Language Model Planners Large language models (LLMs) exhibit a wide range of promising capabilities --- from step-by-step planning to commonsense reasoning --- that may provide utility for robots, but remain prone to confidently hallucinated predictions. In this...
[ -0.05144898220896721, -0.007469890173524618, -0.020442938432097435, 0.0040004076436161995, -0.03379455953836441, -0.0021814110223203897, 0.02380392700433731, -0.03045203536748886, -0.031116846948862076, -0.0028946977108716965, -0.02025826834142208, -0.010627741925418377, -0.03192939236760139...
corl_2023_4uFVn6WHyzo
4uFVn6WHyzo
corl
2,023
Generating Transferable Adversarial Simulation Scenarios for Self-Driving via Neural Rendering
Self-driving software pipelines include components that are learned from a significant number of training examples, yet it remains challenging to evaluate the overall system's safety and generalization performance. Together with scaling up the real-world deployment of autonomous vehicles, it is of critical importance t...
Yasasa Abeysirigoonawardena;Kevin Xie;Chuhan Chen;Salar Hosseini Khorasgani;Ruiting Chen;Ruiqi Wang;Florian Shkurti
Department of Computer Science;Department of Computer Science, University of Toronto;Flawless AI Inc.;Toronto University;;Stanford University;University of Toronto
Poster
main
robotics;adversarial attacks;simulation
https://openreview.net/forum?id=4uFVn6WHyzo
3
Generating Transferable Adversarial Simulation Scenarios for Self-Driving via Neural Rendering Self-driving software pipelines include components that are learned from a significant number of training examples, yet it remains challenging to evaluate the overall system's safety and generalization performance. Together w...
[ -0.06727548688650131, -0.0108369505032897, -0.0033548532519489527, 0.021692683920264244, -0.03470829501748085, -0.009325036779046059, 0.0001719685096759349, 0.02052822709083557, 0.04252141714096069, 0.04999646916985512, 0.003476933343335986, -0.006578235886991024, 0.006639275699853897, 0.0...
corl_2023_4x2RUQ99sGz
4x2RUQ99sGz
corl
2,023
Online Model Adaptation with Feedforward Compensation
To cope with distribution shifts or non-stationarity in system dynamics, online adaptation algorithms have been introduced to update offline-learned prediction models in real-time. Existing online adaptation methods focus on optimizing the prediction model by utilizing feedback from the latest prediction error. Unfort...
ABULIKEMU ABUDUWEILI;Changliu Liu
Carnegie Mellon University;Carnegie Mellon University
Poster
main
Online Adaptation;Optimization;Behavior prediction
https://github.com/intelligent-control-lab/Feedforward_Adaptation
https://openreview.net/forum?id=4x2RUQ99sGz
3
Online Model Adaptation with Feedforward Compensation To cope with distribution shifts or non-stationarity in system dynamics, online adaptation algorithms have been introduced to update offline-learned prediction models in real-time. Existing online adaptation methods focus on optimizing the prediction model by utili...
[ -0.050144582986831665, -0.02870972268283367, -0.015050798654556274, 0.0011332174763083458, -0.05077556520700455, -0.045690588653087616, 0.04806605353951454, 0.023420602083206177, 0.04257278889417648, 0.02629714086651802, -0.04520807042717934, -0.013361991383135319, -0.01847480796277523, 0....
corl_2023_5JMGq83yf1N
5JMGq83yf1N
corl
2,023
Ready, Set, Plan! Planning to Goal Sets Using Generalized Bayesian Inference
Many robotic tasks can have multiple and diverse solutions and, as such, are naturally expressed as goal sets. Examples include navigating to a room, finding a feasible placement location for an object, or opening a drawer enough to reach inside. Using a goal set as a planning objective requires that a model for the ob...
Jana Pavlasek;Stanley Robert Lewis;Balakumar Sundaralingam;Fabio Ramos;Tucker Hermans
University of Michigan;University of Michigan;NVIDIA;NVIDIA;University of Utah
Poster
main
Planning as inference;Variational inference;Nonparametric learning
https://openreview.net/forum?id=5JMGq83yf1N
4
Ready, Set, Plan! Planning to Goal Sets Using Generalized Bayesian Inference Many robotic tasks can have multiple and diverse solutions and, as such, are naturally expressed as goal sets. Examples include navigating to a room, finding a feasible placement location for an object, or opening a drawer enough to reach insi...
[ -0.025491124019026756, -0.007643529679626226, -0.045651767402887344, 0.03485754132270813, -0.006858235690742731, -0.045613691210746765, 0.013611765578389168, -0.008562086150050163, 0.016219893470406532, 0.050449199974536896, -0.0196656696498394, 0.01976085640490055, 0.011765134520828724, 0...
corl_2023_69y5fzvaAT
69y5fzvaAT
corl
2,023
RoboCook: Long-Horizon Elasto-Plastic Object Manipulation with Diverse Tools
Humans excel in complex long-horizon soft body manipulation tasks via flexible tool use: bread baking requires a knife to slice the dough and a rolling pin to flatten it. Often regarded as a hallmark of human cognition, tool use in autonomous robots remains limited due to challenges in understanding tool-object interac...
Haochen Shi;Huazhe Xu;Samuel Clarke;Yunzhu Li;Jiajun Wu
Stanford University;Tsinghua University;;Stanford University;Stanford University
Oral
main
Deformable Object Manipulation;Long-horizon Planning;Model Learning;Tool Usage
https://github.com/hshi74/robocook
https://openreview.net/forum?id=69y5fzvaAT
62
RoboCook: Long-Horizon Elasto-Plastic Object Manipulation with Diverse Tools Humans excel in complex long-horizon soft body manipulation tasks via flexible tool use: bread baking requires a knife to slice the dough and a rolling pin to flatten it. Often regarded as a hallmark of human cognition, tool use in autonomous ...
[ -0.05169619247317314, -0.011118347756564617, 0.026269730180501938, -0.0007653602515347302, -0.015279707498848438, -0.01747954450547695, -0.0010844513308256865, -0.016892921179533005, -0.02311662957072258, 0.010980857536196709, -0.015197213739156723, 0.0012626153184100986, 0.00863894633948803...
corl_2023_6Um8P8Fvyhl
6Um8P8Fvyhl
corl
2,023
MOTO: Offline Pre-training to Online Fine-tuning for Model-based Robot Learning
We study the problem of offline pre-training and online fine-tuning for reinforcement learning from high-dimensional observations in the context of realistic robot tasks. Recent offline model-free approaches successfully use online fine-tuning to either improve the performance of the agent over the data collection poli...
Rafael Rafailov;Kyle Beltran Hatch;Victor Kolev;John D Martin;Mariano Phielipp;Chelsea Finn
Stanford University;Toyota Research Institute;Stanford University;Intel;Intel Labs;Google
Poster
main
offline RL;online fine-tuning;model-learning;robot learning
https://openreview.net/forum?id=6Um8P8Fvyhl
16
MOTO: Offline Pre-training to Online Fine-tuning for Model-based Robot Learning We study the problem of offline pre-training and online fine-tuning for reinforcement learning from high-dimensional observations in the context of realistic robot tasks. Recent offline model-free approaches successfully use online fine-tun...
[ -0.04930432140827179, -0.038574691861867905, 0.004234769847244024, 0.006460054311901331, -0.021422134712338448, -0.002406277460977435, 0.048598915338516235, 0.03372965008020401, 0.007856948301196098, 0.011991939507424831, -0.05216308310627937, -0.0013179995585232973, 0.008218933828175068, ...
corl_2023_6a4sECAMCA
6a4sECAMCA
corl
2,023
Learning to Drive Anywhere
Human drivers can seamlessly adapt their driving decisions across geographical locations with diverse conditions and rules of the road, e.g., left vs. right-hand traffic. In contrast, existing models for autonomous driving have been thus far only deployed within restricted operational domains, \ie, without accounting f...
Ruizhao Zhu;Peng Huang;Eshed Ohn-Bar;Venkatesh Saligrama
Boston University;Boston University;Boston University;Boston University
Poster
main
Global-scale Autonomous Driving;Imitation Learning;Transformer
https://openreview.net/forum?id=6a4sECAMCA
9
Learning to Drive Anywhere Human drivers can seamlessly adapt their driving decisions across geographical locations with diverse conditions and rules of the road, e.g., left vs. right-hand traffic. In contrast, existing models for autonomous driving have been thus far only deployed within restricted operational domains...
[ -0.09609374403953552, -0.03323179855942726, 0.04492790997028351, -0.008363645523786545, -0.015362746082246304, -0.009653930552303791, -0.017711250111460686, 0.018723055720329285, 0.0056020645424723625, 0.02989005483686924, -0.029110314324498177, -0.026418352499604225, -0.01791546866297722, ...
corl_2023_6kSohKYYTn0
6kSohKYYTn0
corl
2,023
Measuring Interpretability of Neural Policies of Robots with Disentangled Representation
The advancement of robots, particularly those functioning in complex human-centric environments, relies on control solutions that are driven by machine learning. Understanding how learning-based controllers make decisions is crucial since robots are mostly safety-critical systems. This urges a formal and quantitative u...
Tsun-Hsuan Wang;Wei Xiao;Tim Seyde;Ramin Hasani;Daniela Rus
Massachusetts Institute of Technology;Massachusetts Institute of Technology;Massachusetts Institute of Technology;Massachusetts Institute of Technology;Massachusetts Institute of Technology
Oral
main
Interpretability;Disentangled Representation;Neural Policy
https://github.com/zswang666/interpret-by-disentangle
https://openreview.net/forum?id=6kSohKYYTn0
8
Measuring Interpretability of Neural Policies of Robots with Disentangled Representation The advancement of robots, particularly those functioning in complex human-centric environments, relies on control solutions that are driven by machine learning. Understanding how learning-based controllers make decisions is crucia...
[ -0.03093082644045353, -0.05240132659673691, -0.009266314096748829, -0.00037156487815082073, -0.026071323081851006, -0.023410607129335403, 0.022856291383504868, -0.009127735160291195, 0.03076453134417534, 0.029341787099838257, -0.032593775540590286, -0.026385435834527016, 0.012084086425602436...
corl_2023_6zGpfOBImD
6zGpfOBImD
corl
2,023
M2T2: Multi-Task Masked Transformer for Object-centric Pick and Place
With the advent of large language models and large-scale robotic datasets, there has been tremendous progress in high-level decision-making for object manipulation. These generic models are able to interpret complex tasks using language commands, but they often have difficulties generalizing to out-of-distribution obje...
Wentao Yuan;Adithyavairavan Murali;Arsalan Mousavian;Dieter Fox
University of Washington, Seattle;;NVIDIA;Department of Computer Science
Poster
main
Object manipulation;Multi-task learning;Pick and place
https://m2-t2.github.io
https://openreview.net/forum?id=6zGpfOBImD
23
M2T2: Multi-Task Masked Transformer for Object-centric Pick and Place With the advent of large language models and large-scale robotic datasets, there has been tremendous progress in high-level decision-making for object manipulation. These generic models are able to interpret complex tasks using language commands, but...
[ -0.004567063879221678, -0.014221718534827232, -0.02353690005838871, 0.03427334874868393, 0.007934651337563992, 0.011967607773840427, -0.004372431896626949, 0.007771703414618969, -0.006545068696141243, 0.004218536429107189, -0.041787050664424896, -0.0259449053555727, 0.020205523818731308, 0...
corl_2023_770xKAHeFS
770xKAHeFS
corl
2,023
How to Learn and Generalize From Three Minutes of Data: Physics-Constrained and Uncertainty-Aware Neural Stochastic Differential Equations
We present a framework and algorithms to learn controlled dynamics models using neural stochastic differential equations (SDEs)---SDEs whose drift and diffusion terms are both parametrized by neural networks. We construct the drift term to leverage a priori physics knowledge as inductive bias, and we design the diffusi...
Franck Djeumou;Cyrus Neary;ufuk topcu
The University of Texas at Austin;University of Texas, Austin;University of Texas, Austin
Oral
main
Neural SDE;Physics-Informed Learning;Data-Driven Modeling;Dynamical Systems;Control;Model-Based Reinforcement Learning
https://github.com/wuwushrek/sde4mbrl
https://openreview.net/forum?id=770xKAHeFS
13
How to Learn and Generalize From Three Minutes of Data: Physics-Constrained and Uncertainty-Aware Neural Stochastic Differential Equations We present a framework and algorithms to learn controlled dynamics models using neural stochastic differential equations (SDEs)---SDEs whose drift and diffusion terms are both param...
[ -0.044682323932647705, 0.0019841257017105818, -0.013512499630451202, -0.018317151814699173, -0.022601375356316566, -0.03695027530193329, -0.0020515022333711386, 0.025593826547265053, 0.009999621659517288, 0.036504194140434265, -0.06129879504442215, 0.002301260596141219, -0.018010471016168594...
corl_2023_7CtUcT_OHmC
7CtUcT_OHmC
corl
2,023
Learning Human Contribution Preferences in Collaborative Human-Robot Tasks
In human-robot collaboration, both human and robotic agents must work together to achieve a set of shared objectives. However, each team member may have individual preferences, or constraints, for how they would like to contribute to the task. Effective teams align their actions to optimize task performance while satis...
Michelle D Zhao;Reid Simmons;Henny Admoni
Carnegie Mellon University;Carnegie Mellon University;Carnegie Mellon University
Poster
main
human-robot collaboration;reward learning;human-robot interaction
https://openreview.net/forum?id=7CtUcT_OHmC
10
Learning Human Contribution Preferences in Collaborative Human-Robot Tasks In human-robot collaboration, both human and robotic agents must work together to achieve a set of shared objectives. However, each team member may have individual preferences, or constraints, for how they would like to contribute to the task. E...
[ -0.021289978176355362, -0.034192994236946106, -0.03765837848186493, 0.017990492284297943, -0.024331403896212578, -0.028478803113102913, 0.02099505253136158, -0.03955696523189545, 0.015013582073152065, 0.014359215274453163, -0.043206676840782166, -0.024331403896212578, 0.0033870420884341, -...
corl_2023_7Pkzm2FgUmq
7Pkzm2FgUmq
corl
2,023
SLAP: Spatial-Language Attention Policies
Despite great strides in language-guided manipulation, existing work has been constrained to table-top settings. Table-tops allow for perfect and consistent camera angles, properties are that do not hold in mobile manipulation. Task plans that involve moving around the environment must be robust to egocentric views and...
Priyam Parashar;Vidhi Jain;Xiaohan Zhang;Jay Vakil;Sam Powers;Yonatan Bisk;Chris Paxton
Meta Facebook;Google;State University of New York at Binghamton;Meta AI ;Carnegie Mellon University;Meta;Meta Platforms
Poster
main
learning from demonstration;language-based robotics
https://github.com/facebookresearch/home-robot/tree/main/projects/slap_manipulation
https://openreview.net/forum?id=7Pkzm2FgUmq
8
SLAP: Spatial-Language Attention Policies Despite great strides in language-guided manipulation, existing work has been constrained to table-top settings. Table-tops allow for perfect and consistent camera angles, properties are that do not hold in mobile manipulation. Task plans that involve moving around the environm...
[ -0.033042795956134796, -0.02355673722922802, -0.01830120012164116, -0.006667265202850103, -0.022960364818572998, 0.04621891677379608, 0.025010397657752037, 0.0003468165814410895, -0.018972119316458702, 0.027414526790380478, -0.030079567804932594, 0.03455236554145813, 0.0012405026936903596, ...
corl_2023_7TYeO2XVqI
7TYeO2XVqI
corl
2,023
SayTap: Language to Quadrupedal Locomotion
Large language models (LLMs) have demonstrated the potential to perform high-level planning. Yet, it remains a challenge for LLMs to comprehend low-level commands, such as joint angle targets or motor torques. This paper proposes an approach to use foot contact patterns as an interface that bridges human commands in na...
Yujin Tang;Wenhao Yu;Jie Tan;Heiga Zen;Aleksandra Faust;Tatsuya Harada
Google;Google;Google;Google;Google Brain;The University of Tokyo
Poster
main
Large language model (LLM);Quadrupedal robots;Locomotion
https://openreview.net/forum?id=7TYeO2XVqI
45
SayTap: Language to Quadrupedal Locomotion Large language models (LLMs) have demonstrated the potential to perform high-level planning. Yet, it remains a challenge for LLMs to comprehend low-level commands, such as joint angle targets or motor torques. This paper proposes an approach to use foot contact patterns as an ...
[ -0.043177131563425064, -0.01848687045276165, -0.0014844511169940233, 0.010291561484336853, -0.022821657359600067, -0.008323360234498978, 0.023656364530324936, -0.013715758919715881, 0.0056485021486878395, -0.0004280246212147176, -0.0017867953283712268, 0.022631950676441193, 0.008788139559328...
corl_2023_86aMPJn6hX9F
86aMPJn6hX9F
corl
2,023
Stabilize to Act: Learning to Coordinate for Bimanual Manipulation
Key to rich, dexterous manipulation in the real world is the ability to coordinate control across two hands. However, while the promise afforded by bimanual robotic systems is immense, constructing control policies for dual arm autonomous systems brings inherent difficulties. One such difficulty is the high-dimensional...
Jennifer Grannen;Yilin Wu;Brandon Vu;Dorsa Sadigh
Computer Science Department, Stanford University;Stanford University;Computer Science Department, Stanford University;Stanford University
Oral
main
Bimanual Manipulation;Learning from Demonstrations;Deformable Object Manipulation
https://openreview.net/forum?id=86aMPJn6hX9F
33
Stabilize to Act: Learning to Coordinate for Bimanual Manipulation Key to rich, dexterous manipulation in the real world is the ability to coordinate control across two hands. However, while the promise afforded by bimanual robotic systems is immense, constructing control policies for dual arm autonomous systems brings...
[ -0.01961592771112919, -0.009760398417711258, -0.039269909262657166, 0.009170589037239552, -0.019340049475431442, -0.026103833690285683, 0.008419057354331017, 0.020643338561058044, 0.020319893956184387, 0.060807473957538605, -0.038432758301496506, -0.023192837834358215, 0.005898096598684788, ...
corl_2023_8L6pHd9aS6w
8L6pHd9aS6w
corl
2,023
XSkill: Cross Embodiment Skill Discovery
Human demonstration videos are a widely available data source for robot learning and an intuitive user interface for expressing desired behavior. However, directly extracting reusable robot manipulation skills from unstructured human videos is challenging due to the big embodiment difference and unobserved action para...
Mengda Xu;Zhenjia Xu;Cheng Chi;Manuela Veloso;Shuran Song
Columbia University;Columbia University;Columbia University;School of Computer Science, Carnegie Mellon University;Columbia University
Poster
main
Manipulation;Representation Learning;Cross-Embodiements
https://xskill.cs.columbia.edu/
https://openreview.net/forum?id=8L6pHd9aS6w
70
XSkill: Cross Embodiment Skill Discovery Human demonstration videos are a widely available data source for robot learning and an intuitive user interface for expressing desired behavior. However, directly extracting reusable robot manipulation skills from unstructured human videos is challenging due to the big embodim...
[ -0.09136214852333069, -0.04366788640618324, -0.026153147220611572, 0.016791822388768196, -0.027159741148352623, -0.017871621996164322, 0.03155215457081795, 0.014311939477920532, -0.005746737588196993, -0.006863142363727093, -0.018704351037740707, -0.046083711087703705, 0.00910510215908289, ...
corl_2023_8asqEWO479I
8asqEWO479I
corl
2,023
Push Past Green: Learning to Look Behind Plant Foliage by Moving It
Autonomous agriculture applications (e.g., inspection, phenotyping, plucking fruits) require manipulating the plant foliage to look behind the leaves and the branches. Partial visibility, extreme clutter, thin structures, and unknown geometry and dynamics for plants make such manipulation challenging. We tackle these ...
Xiaoyu Zhang;Saurabh Gupta
Department of Computer Science;University of Illinois, Urbana Champaign
Poster
main
Deformable Object Manipulation;Model-building;Self-supervision
https://github.com/ErinZhang1998/pushpastgreen
https://openreview.net/forum?id=8asqEWO479I
3
Push Past Green: Learning to Look Behind Plant Foliage by Moving It Autonomous agriculture applications (e.g., inspection, phenotyping, plucking fruits) require manipulating the plant foliage to look behind the leaves and the branches. Partial visibility, extreme clutter, thin structures, and unknown geometry and dyna...
[ -0.056466273963451385, -0.001769205555319786, -0.03696446120738983, 0.013115525245666504, -0.03605610877275467, -0.051016148179769516, 0.01236474234610796, -0.005848689936101437, -0.00028024008497595787, 0.03370180353522301, -0.023821130394935608, 0.010974404402077198, -0.026972563937306404,...
corl_2023_8scj3Y0RLq
8scj3Y0RLq
corl
2,023
Task Generalization with Stability Guarantees via Elastic Dynamical System Motion Policies
Dynamical System (DS) based Learning from Demonstration (LfD) allows learning of reactive motion policies with stability and convergence guarantees from a few trajectories. Yet, current DS learning techniques lack the flexibility to generalize to new task instances as they overlook explicit task parameters that inheren...
Tianyu Li;Nadia Figueroa
University of Pennsylvania;University of Pennsylvania
Poster
main
Stable Dynamical Systems;Reactive Motion Policies;Learning from Demonstrations;Task Parametrization;Task Generalization
https://openreview.net/forum?id=8scj3Y0RLq
7
Task Generalization with Stability Guarantees via Elastic Dynamical System Motion Policies Dynamical System (DS) based Learning from Demonstration (LfD) allows learning of reactive motion policies with stability and convergence guarantees from a few trajectories. Yet, current DS learning techniques lack the flexibility...
[ -0.03658170625567436, -0.03422752767801285, -0.016074616461992264, -0.0019771410152316093, -0.016102204099297523, 0.0029795055743306875, -0.0020127755124121904, -0.0034530996344983578, 0.0012610021512955427, 0.031303200870752335, -0.03490803390741348, 0.016994217410683632, 0.0271649975329637...
corl_2023_8yTS_nAILxt
8yTS_nAILxt
corl
2,023
REFLECT: Summarizing Robot Experiences for Failure Explanation and Correction
The ability to detect and analyze failed executions automatically is crucial for an explainable and robust robotic system. Recently, Large Language Models (LLMs) have demonstrated strong reasoning abilities on textual inputs. To leverage the power of LLMs for robot failure explanation, we introduce REFLECT, a framework...
Zeyi Liu;Arpit Bahety;Shuran Song
Columbia University;Columbia University;Columbia University
Poster
main
Large Language Model;Explainable AI;Task Planning
https://github.com/real-stanford/reflect
https://openreview.net/forum?id=8yTS_nAILxt
136
REFLECT: Summarizing Robot Experiences for Failure Explanation and Correction The ability to detect and analyze failed executions automatically is crucial for an explainable and robust robotic system. Recently, Large Language Models (LLMs) have demonstrated strong reasoning abilities on textual inputs. To leverage the ...
[ -0.09858878701925278, -0.024052638560533524, 0.027331724762916565, -0.01701701618731022, -0.012593853287398815, -0.0037565354723483324, 0.031619761139154434, -0.003911931999027729, -0.006914007477462292, -0.0024773315526545048, -0.010350742377340794, -0.01824216917157173, -0.0023444565013051...
corl_2023_93qz1k6_6h
93qz1k6_6h
corl
2,023
Dexterous Functional Grasping
While there have been significant strides in dexterous manipulation, most of it is limited to benchmark tasks like in-hand reorientation which are of limited utility in the real world. The main benefit of dexterous hands over two-fingered ones is their ability to pickup tools and other objects (including thin ones) and...
Ananye Agarwal;Shagun Uppal;Kenneth Shaw;Deepak Pathak
Carnegie Mellon University;Carnegie Mellon University;Carnegie Mellon University;Carnegie Mellon University
Poster
main
Functional Grasping;Sim2real
https://openreview.net/forum?id=93qz1k6_6h
34
Dexterous Functional Grasping While there have been significant strides in dexterous manipulation, most of it is limited to benchmark tasks like in-hand reorientation which are of limited utility in the real world. The main benefit of dexterous hands over two-fingered ones is their ability to pickup tools and other obj...
[ -0.025028353556990623, -0.012190123088657856, -0.0276398416608572, 0.016498126089572906, -0.032767508178949356, -0.030441949144005775, 0.03328217938542366, -0.00910208560526371, 0.006485831458121538, 0.04422374442219734, -0.04269878938794136, -0.012914477847516537, 0.013362433761358261, -0...
corl_2023_9GRE34K0SB
9GRE34K0SB
corl
2,023
AdaptSim: Task-Driven Simulation Adaptation for Sim-to-Real Transfer
Simulation parameter settings such as contact models and object geometry approximations are critical to training robust manipulation policies capable of transferring from simulation to real-world deployment. There is often an irreducible gap between simulation and reality: attempting to match the dynamics between simul...
Allen Z. Ren;Hongkai Dai;Benjamin Burchfiel;Anirudha Majumdar
Google DeepMind;Toyota Research Institute;Dexterous Manipulation Group, Toyota Research Institute;Princeton University
Poster
main
Contact-rich manipulation;sim-to-real transfer
https://github.com/irom-lab/AdaptSim
https://openreview.net/forum?id=9GRE34K0SB
16
AdaptSim: Task-Driven Simulation Adaptation for Sim-to-Real Transfer Simulation parameter settings such as contact models and object geometry approximations are critical to training robust manipulation policies capable of transferring from simulation to real-world deployment. There is often an irreducible gap between s...
[ -0.026649193838238716, -0.01772630773484707, -0.02247721701860428, 0.005688224453479052, -0.05859147012233734, -0.04091110825538635, 0.01034723874181509, 0.006262560375034809, 0.03589370846748352, 0.0417565293610096, -0.03826456516981125, 0.02348804846405983, 0.002867085626348853, 0.036959...
corl_2023_9SM6l0HyY_
9SM6l0HyY_
corl
2,023
Learning Generalizable Manipulation Policies with Object-Centric 3D Representations
We introduce GROOT, an imitation learning method for learning robust policies with object-centric and 3D priors. GROOT builds policies that generalize beyond their initial training conditions for vision-based manipulation. It constructs object-centric 3D representations that are robust toward background changes and cam...
Yifeng Zhu;Zhenyu Jiang;Peter Stone;Yuke Zhu
The University of Texas at Austin;University of Texas, Austin;University of Texas, Austin;Computer Science Department, University of Texas, Austin
Poster
main
robot manipulation;imitation learning;object-centric representations
https://github.com/UT-Austin-RPL/GROOT
https://openreview.net/forum?id=9SM6l0HyY_
48
Learning Generalizable Manipulation Policies with Object-Centric 3D Representations We introduce GROOT, an imitation learning method for learning robust policies with object-centric and 3D priors. GROOT builds policies that generalize beyond their initial training conditions for vision-based manipulation. It constructs...
[ -0.046896062791347504, -0.009009788744151592, 0.012112407013773918, -0.013160161674022675, -0.025073856115341187, -0.04844962805509567, 0.001390984863974154, -0.012654349207878113, 0.004288117401301861, -0.0028768095653504133, -0.013909848406910896, -0.0009004143066704273, 0.0048142527230083...
corl_2023_9_8LF30mOC
9_8LF30mOC
corl
2,023
VoxPoser: Composable 3D Value Maps for Robotic Manipulation with Language Models
Large language models (LLMs) are shown to possess a wealth of actionable knowledge that can be extracted for robot manipulation in the form of reasoning and planning. Despite the progress, most still rely on pre-defined motion primitives to carry out the physical interactions with the environment, which remains a major...
Wenlong Huang;Chen Wang;Ruohan Zhang;Yunzhu Li;Jiajun Wu;Li Fei-Fei
Stanford University;Computer Science Department, Stanford University;Stanford University;Stanford University;Stanford University;Stanford University
Oral
main
Manipulation;Large Language Models;Model-based Planning
https://github.com/huangwl18/VoxPoser
https://openreview.net/forum?id=9_8LF30mOC
564
VoxPoser: Composable 3D Value Maps for Robotic Manipulation with Language Models Large language models (LLMs) are shown to possess a wealth of actionable knowledge that can be extracted for robot manipulation in the form of reasoning and planning. Despite the progress, most still rely on pre-defined motion primitives t...
[ -0.03790971264243126, -0.027379237115383148, 0.012975377961993217, -0.0025090002454817295, -0.019174622371792793, 0.014532056637108326, 0.02933882176876068, -0.01153773907572031, 0.014229877851903439, -0.008277869783341885, 0.0034544540103524923, -0.006327442359179258, 0.0365544855594635, ...
corl_2023_9al6taqfTzr
9al6taqfTzr
corl
2,023
Open-World Object Manipulation using Pre-Trained Vision-Language Models
For robots to follow instructions from people, they must be able to connect the rich semantic information in human vocabulary, e.g. ``can you get me the pink stuffed whale?'' to their sensory observations and actions. This brings up a notably difficult challenge for robots: while robot learning approaches allow robots ...
Austin Stone;Ted Xiao;Yao Lu;Keerthana Gopalakrishnan;Kuang-Huei Lee;Quan Vuong;Paul Wohlhart;Sean Kirmani;Brianna Zitkovich;Fei Xia;Chelsea Finn;Karol Hausman
Google;;Google;Research, Google;Google;;Graz University of Technology;Google X;;Google;Google;
Poster
main
https://openreview.net/forum?id=9al6taqfTzr
161
Open-World Object Manipulation using Pre-Trained Vision-Language Models For robots to follow instructions from people, they must be able to connect the rich semantic information in human vocabulary, e.g. ``can you get me the pink stuffed whale?'' to their sensory observations and actions. This brings up a notably diffi...
[ -0.04602865129709244, -0.026410795748233795, -0.017836103215813637, 0.022179128602147102, -0.003020630218088627, -0.021603770554065704, 0.0026819114573299885, 0.016843145713210106, -0.01956217736005783, 0.027784230187535286, -0.011117404326796532, -0.048292964696884155, 0.01789178140461445, ...
corl_2023_9bK38pUBzU
9bK38pUBzU
corl
2,023
Language-Conditioned Path Planning
Contact is at the core of robotic manipulation. At times, it is desired (e.g. manipulation and grasping), and at times, it is harmful (e.g. when avoiding obstacles). However, traditional path planning algorithms focus solely on collision-free paths, limiting their applicability in contact-rich tasks. To address this li...
Amber Xie;Youngwoon Lee;Pieter Abbeel;Stephen James
;University of California, Berkeley;Covariant;Dyson
Poster
main
Robotic Manipulation;Path Planning;Collision Avoidance;Learned Collision Function
https://github.com/amberxie88/lapp
https://openreview.net/forum?id=9bK38pUBzU
16
Language-Conditioned Path Planning Contact is at the core of robotic manipulation. At times, it is desired (e.g. manipulation and grasping), and at times, it is harmful (e.g. when avoiding obstacles). However, traditional path planning algorithms focus solely on collision-free paths, limiting their applicability in con...
[ -0.022052101790905, -0.02245740406215191, 0.0015797558007761836, 0.002381148049607873, -0.015069857239723206, 0.014977742917835712, 0.023378543555736542, 0.0242259930819273, -0.011293181218206882, 0.034542765468358994, -0.016000209376215935, -0.028039515018463135, 0.041119709610939026, 0.0...
corl_2023_9cTEQWMo1BF
9cTEQWMo1BF
corl
2,023
LabelFormer: Object Trajectory Refinement for Offboard Perception from LiDAR Point Clouds
A major bottleneck to scaling-up training of self-driving perception systems are the human annotations required for supervision. A promising alternative is to leverage “auto-labelling” offboard perception models that are trained to automatically generate annotations from raw LiDAR point clouds at a fraction of the cost...
Anqi Joyce Yang;Sergio Casas;Nikita Dvornik;Sean Segal;Yuwen Xiong;Jordan Sir Kwang Hu;Carter Fang;Raquel Urtasun
Waabi Innovation Inc;Waabi;Department of Computer Science, University of Toronto;Department of Computer Science, University of Toronto;Waabi Innovation Inc.;;Department of Computer Science, University of Toronto;University of Toronto
Poster
main
Auto-labelling;Offboard Perception;Trajectory Refinement;Transformer
https://openreview.net/forum?id=9cTEQWMo1BF
8
LabelFormer: Object Trajectory Refinement for Offboard Perception from LiDAR Point Clouds A major bottleneck to scaling-up training of self-driving perception systems are the human annotations required for supervision. A promising alternative is to leverage “auto-labelling” offboard perception models that are trained to...
[ -0.07884375005960464, -0.031221242621541023, -0.05516120791435242, 0.011565466411411762, -0.011372402310371399, 0.015316428616642952, 0.014672880992293358, 0.029364148154854774, 0.0049828956834971905, 0.02675318531692028, -0.02987898699939251, -0.03727058693766594, 0.014112075790762901, 0....
corl_2023_AIgm8ZE_DlD
AIgm8ZE_DlD
corl
2,023
A Universal Semantic-Geometric Representation for Robotic Manipulation
Robots rely heavily on sensors, especially RGB and depth cameras, to perceive and interact with the world. RGB cameras record 2D images with rich semantic information while missing precise spatial information. On the other side, depth cameras offer critical 3D geometry data but capture limited semantics. Therefore, int...
Tong Zhang;Yingdong Hu;Hanchen Cui;Hang Zhao;Yang Gao
Tsinghua University;Tsinghua University;Shanghai Qi Zhi Institute;Tsinghua University;Tsinghua University
Poster
main
Representation Learning;Robotic Manipulation
https://openreview.net/forum?id=AIgm8ZE_DlD
22
A Universal Semantic-Geometric Representation for Robotic Manipulation Robots rely heavily on sensors, especially RGB and depth cameras, to perceive and interact with the world. RGB cameras record 2D images with rich semantic information while missing precise spatial information. On the other side, depth cameras offer ...
[ -0.09454251825809479, 0.007621173281222582, -0.0036516229156404734, -0.006862689275294542, -0.006517511326819658, -0.06482084840536118, -0.011726978234946728, -0.01759500987827778, 0.026306219398975372, 0.014006971381604671, -0.00525488518178463, 0.027468925341963768, 0.021146712824702263, ...
corl_2023_ANJuNDFdvP
ANJuNDFdvP
corl
2,023
UniFolding: Towards Sample-efficient, Scalable, and Generalizable Robotic Garment Folding
This paper explores the development of UniFolding, a sample-efficient, scalable, and generalizable robotic system for unfolding and folding various garments. UniFolding employs the proposed UFONet neural network to integrate unfolding and folding decisions into a single policy model that is adaptable to different garm...
Han Xue;Yutong Li;Wenqiang Xu;Huanyu Li;Dongzhe Zheng;Cewu Lu
Shanghai Jiaotong University;Shanghai Jiaotong University;Shanghai Jiaotong University;;Shanghai Jiaotong University;Shanghai Jiaotong University
Poster
main
Deformable Object Manipulation;Bimanual Manipulation;Garment Folding
https://github.com/xiaoxiaoxh/UniFolding
https://openreview.net/forum?id=ANJuNDFdvP
15
UniFolding: Towards Sample-efficient, Scalable, and Generalizable Robotic Garment Folding This paper explores the development of UniFolding, a sample-efficient, scalable, and generalizable robotic system for unfolding and folding various garments. UniFolding employs the proposed UFONet neural network to integrate unfo...
[ -0.07040730118751526, -0.015412080101668835, -0.07886810600757599, 0.021673819050192833, -0.014564136043190956, -0.008134670555591583, 0.013529830612242222, 0.030842795968055725, -0.015002084895968437, 0.024692872539162636, -0.031234154477715492, 0.005143572110682726, 0.005837767850607634, ...
corl_2023_AnDDMQgM7-
AnDDMQgM7-
corl
2,023
Equivariant Reinforcement Learning under Partial Observability
Incorporating inductive biases is a promising approach for tackling challenging robot learning domains with sample-efficient solutions. This paper identifies partially observable domains where symmetries can be a useful inductive bias for efficient learning. Specifically, by encoding the equivariance regarding specific...
Hai Huu Nguyen;Andrea Baisero;David Klee;Dian Wang;Robert Platt;Christopher Amato
Northeastern University;Northeastern University;Boston Dynamics Artificial Intelligence Institute;Boston Dynamics AI Institute;Northeastern University;Northeastern University
Poster
main
Partial Observability;Equivariant Learning;Symmetry
https://github.com/hai-h-nguyen/equi-rl-for-pomdps
https://openreview.net/forum?id=AnDDMQgM7-
15
Equivariant Reinforcement Learning under Partial Observability Incorporating inductive biases is a promising approach for tackling challenging robot learning domains with sample-efficient solutions. This paper identifies partially observable domains where symmetries can be a useful inductive bias for efficient learning...
[ -0.058721497654914856, 0.014851076528429985, -0.0040636565536260605, 0.015666652470827103, -0.004511749371886253, -0.04138575866818428, 0.05162787809967995, 0.013504426926374435, 0.04294104129076004, 0.04946565255522728, -0.04191683232784271, -0.01151290349662304, -0.007264319341629744, 0....
corl_2023_ApxLUk8U-l
ApxLUk8U-l
corl
2,023
Self-Improving Robots: End-to-End Autonomous Visuomotor Reinforcement Learning
In imitation and reinforcement learning (RL), the cost of human supervision limits the amount of data that the robots can be trained on. While RL offers a framework for building self-improving robots that can learn via trial-and-error autonomously, practical realizations end up requiring extensive human supervision for...
Archit Sharma;Ahmed M Ahmed;Rehaan Ahmad;Chelsea Finn
Stanford University;;Computer Science Department, Stanford University;Google
Poster
main
reinforcement learning;autonomous;reset-free;manipulation
https://github.com/rehaanahmad2013/self-improving-robots/
https://openreview.net/forum?id=ApxLUk8U-l
16
Self-Improving Robots: End-to-End Autonomous Visuomotor Reinforcement Learning In imitation and reinforcement learning (RL), the cost of human supervision limits the amount of data that the robots can be trained on. While RL offers a framework for building self-improving robots that can learn via trial-and-error autono...
[ -0.08630195260047913, 0.002030264586210251, 0.027997871860861778, -0.004821442067623138, -0.024684295058250427, -0.0056265671737492085, 0.00840029213577509, 0.04825165122747421, 0.006333960220217705, 0.00954049825668335, -0.03213053569197655, -0.011262441985309124, -0.01062020380049944, 0....
corl_2023_AyRr_i028w
AyRr_i028w
corl
2,023
Cold Diffusion on the Replay Buffer: Learning to Plan from Known Good States
Learning from demonstrations (LfD) has successfully trained robots to exhibit remarkable generalization capabilities. However, many powerful imitation techniques do not prioritize the feasibility of the robot behaviors they generate. In this work, we explore the feasibility of plans produced by LfD. As in prior work, w...
Zidan Wang;Takeru Oba;Takuma Yoneda;Rui Shen;Matthew Walter;Bradly C. Stadie
Northwestern University;Toyota Technological Institute;Toyota Technological Institute at Chicago;Yale University;Toyota Technological Institute at Chicago;Northwestern University
Poster
main
Imitation Learning;Reinforcement Learning;Diffusion;Cold Diffusion;Planning;Safety
https://github.com/zidanwang2025/cold_diffusion_on_replay_buffer
https://openreview.net/forum?id=AyRr_i028w
6
Cold Diffusion on the Replay Buffer: Learning to Plan from Known Good States Learning from demonstrations (LfD) has successfully trained robots to exhibit remarkable generalization capabilities. However, many powerful imitation techniques do not prioritize the feasibility of the robot behaviors they generate. In this w...
[ -0.08669732511043549, -0.02681739255785942, -0.014014842920005322, -0.006874252576380968, -0.023584609851241112, 0.007737552281469107, 0.01118615921586752, 0.0022971576545387506, 0.014859774149954319, 0.02819499745965004, -0.027129650115966797, -0.03451361507177353, -0.0020273765549063683, ...
corl_2023_B7PnAw4ze0l
B7PnAw4ze0l
corl
2,023
Precise Robotic Needle-Threading with Tactile Perception and Reinforcement Learning
This work presents a novel tactile perception-based method, named T-NT, for performing the needle-threading task, an application of deformable linear object (DLO) manipulation. This task is divided into two main stages: \textit{Tail-end Finding} and \textit{Tail-end Insertion}. In the first stage, the agent traces the ...
Zhenjun Yu;Wenqiang Xu;Siqiong Yao;Jieji Ren;Tutian Tang;Yutong Li;Guoying Gu;Cewu Lu
Shanghai Jiaotong University;Shanghai Jiaotong University;Shanghai Jiaotong University;Shanghai Jiaotong University;Shanghai Jiaotong University;Shanghai Jiaotong University;Shanghai Jiaotong University;Shanghai Jiaotong University
Poster
main
tactile perception;needle threading
https://openreview.net/forum?id=B7PnAw4ze0l
8
Precise Robotic Needle-Threading with Tactile Perception and Reinforcement Learning This work presents a novel tactile perception-based method, named T-NT, for performing the needle-threading task, an application of deformable linear object (DLO) manipulation. This task is divided into two main stages: \textit{Tail-end...
[ -0.024706970900297165, -0.017327677458524704, -0.012171280570328236, 0.04551475867629051, 0.007260861340910196, 0.018603110685944557, 0.029936248436570168, 0.01029457151889801, 0.005256608594208956, 0.02073490619659424, -0.006085640285164118, -0.027749791741371155, 0.015214101411402225, 0....
corl_2023_BimpCf1rT7
BimpCf1rT7
corl
2,023
Compositional Diffusion-Based Continuous Constraint Solvers
This paper introduces an approach for learning to solve continuous constraint satisfaction problems (CCSP) in robotic reasoning and planning. Previous methods primarily rely on hand-engineering or learning generators for specific constraint types and then rejecting the value assignments when other constraints are viola...
Zhutian Yang;Jiayuan Mao;Yilun Du;Jiajun Wu;Joshua B. Tenenbaum;Tomás Lozano-Pérez;Leslie Pack Kaelbling
Massachusetts Institute of Technology;Massachusetts Institute of Technology;Massachusetts Institute of Technology;Stanford University;Massachusetts Institute of Technology;Massachusetts Institute of Technology;Massachusetts Institute of Technology
Poster
main
Diffusion Models;Constraint Satisfaction Problems;Task and Motion Planning
https://openreview.net/forum?id=BimpCf1rT7
29
Compositional Diffusion-Based Continuous Constraint Solvers This paper introduces an approach for learning to solve continuous constraint satisfaction problems (CCSP) in robotic reasoning and planning. Previous methods primarily rely on hand-engineering or learning generators for specific constraint types and then reje...
[ -0.06369047611951828, -0.010375688783824444, -0.034472234547138214, 0.02944503352046013, -0.010640278458595276, -0.038970254361629486, 0.01321057602763176, 0.008665306493639946, 0.03838437795639038, 0.04161614924669266, -0.0019548912532627583, -0.025513987988233566, -0.0011475388891994953, ...
corl_2023_BzjLaVvr955
BzjLaVvr955
corl
2,023
Topology-Matching Normalizing Flows for Out-of-Distribution Detection in Robot Learning
To facilitate reliable deployments of autonomous robots in the real world, Out-of-Distribution (OOD) detection capabilities are often required. A powerful approach for OOD detection is based on density estimation with Normalizing Flows (NFs). However, we find that prior work with NFs attempts to match the complex targe...
Jianxiang Feng;Jongseok Lee;Simon Geisler;Stephan Günnemann;Rudolph Triebel
RMC, German Aerospace Center (DLR);German Aerospace Center (DLR);Technical University Munich;Technical University Munich;Technical University Munich
Poster
main
Normalizing Flows;Out-of-Distribution Detection;Robotic Introspection
https://github.com/DLR-RM
https://openreview.net/forum?id=BzjLaVvr955
6
Topology-Matching Normalizing Flows for Out-of-Distribution Detection in Robot Learning To facilitate reliable deployments of autonomous robots in the real world, Out-of-Distribution (OOD) detection capabilities are often required. A powerful approach for OOD detection is based on density estimation with Normalizing Fl...
[ -0.02578311413526535, -0.038546670228242874, -0.01232469454407692, 0.005686914082616568, -0.012973844073712826, -0.03883924335241318, -0.010194387286901474, 0.0027977421414107084, 0.01839561201632023, 0.03620607405900955, -0.037486087530851364, -0.039534106850624084, -0.005097193643450737, ...
corl_2023_C5MQUlzhVjQ
C5MQUlzhVjQ
corl
2,023
Surrogate Assisted Generation of Human-Robot Interaction Scenarios
As human-robot interaction (HRI) systems advance, so does the difficulty of evaluating and understanding the strengths and limitations of these systems in different environments and with different users. To this end, previous methods have algorithmically generated diverse scenarios that reveal system failures in a shar...
Varun Bhatt;Heramb Nemlekar;Matthew Christopher Fontaine;Bryon Tjanaka;Hejia Zhang;Ya-Chuan Hsu;Stefanos Nikolaidis
University of Southern California;University of Southern California;University of Southern California;InstaDeep;University of Southern California;University of Southern California;University of Southern California
Oral
main
Scenario Generation;Human-Robot Interaction;Quality Diversity
https://github.com/icaros-usc/dsas
https://openreview.net/forum?id=C5MQUlzhVjQ
11
Surrogate Assisted Generation of Human-Robot Interaction Scenarios As human-robot interaction (HRI) systems advance, so does the difficulty of evaluating and understanding the strengths and limitations of these systems in different environments and with different users. To this end, previous methods have algorithmicall...
[ -0.036638736724853516, 0.005821634083986282, -0.0036643431521952152, 0.013436706736683846, -0.008596300147473812, -0.0018368663731962442, -0.006385018117725849, 0.03250725194811821, -0.006070462055504322, 0.027474356815218925, -0.028244314715266228, -0.013981311582028866, -0.0077840881422162...
corl_2023_CnKf9TyYtf2
CnKf9TyYtf2
corl
2,023
Few-Shot In-Context Imitation Learning via Implicit Graph Alignment
Consider the following problem: given a few demonstrations of a task across a few different objects, how can a robot learn to perform that same task on new, previously unseen objects? This is challenging because the large variety of objects within a class makes it difficult to infer the task-relevant relationship betwe...
Vitalis Vosylius;Edward Johns
Imperial College London;Imperial College London
Poster
main
Few-Shot Imitation Learning;Graph Neural Networks
https://openreview.net/forum?id=CnKf9TyYtf2
11
Few-Shot In-Context Imitation Learning via Implicit Graph Alignment Consider the following problem: given a few demonstrations of a task across a few different objects, how can a robot learn to perform that same task on new, previously unseen objects? This is challenging because the large variety of objects within a cl...
[ -0.059670183807611465, -0.03688735514879227, 0.006166162434965372, 0.03495072200894356, -0.006755373906344175, -0.020243283361196518, 0.002662868704646826, 0.007591231260448694, 0.0065726726315915585, -0.01077936589717865, -0.05009664595127106, -0.028117701411247253, 0.006842156872153282, ...
corl_2023_D0X97ODIYK
D0X97ODIYK
corl
2,023
Structural Concept Learning via Graph Attention for Multi-Level Rearrangement Planning
Robotic manipulation tasks, such as object rearrangement, play a crucial role in enabling robots to interact with complex and arbitrary environments. Existing work focuses primarily on single-level rearrangement planning and, even if multiple levels exist, dependency relations among substructures are geometrically simp...
Manav Kulshrestha;Ahmed H Qureshi
Purdue University;Purdue University
Poster
main
Rearrangement Planning;Robot Manipulation;Graph Attention
https://openreview.net/forum?id=D0X97ODIYK
6
Structural Concept Learning via Graph Attention for Multi-Level Rearrangement Planning Robotic manipulation tasks, such as object rearrangement, play a crucial role in enabling robots to interact with complex and arbitrary environments. Existing work focuses primarily on single-level rearrangement planning and, even if...
[ -0.07575945556163788, -0.011493969708681107, -0.009369187988340855, -0.0011528314789757133, -0.00716197956353426, -0.0002012017066590488, 0.028886040672659874, -0.01636631414294243, 0.012446458451449871, 0.001197479316033423, -0.008297638967633247, -0.017822522670030594, -0.03086428530514240...
corl_2023_DYPOvNot5F
DYPOvNot5F
corl
2,023
Diff-LfD: Contact-aware Model-based Learning from Visual Demonstration for Robotic Manipulation via Differentiable Physics-based Simulation and Rendering
Learning from Demonstration (LfD) is an efficient technique for robots to acquire new skills through expert observation, significantly mitigating the need for laborious manual reward function design. This paper introduces a novel framework for model-based LfD in the context of robotic manipulation. Our proposed pipelin...
Xinghao Zhu;Jinghan Ke;Zhixuan Xu;Zhixin Sun;Bizhe Bai;Jun Lv;Qingtao Liu;Yuwei Zeng;Qi Ye;Cewu Lu;Masayoshi Tomizuka;Lin Shao
;;Zhejiang University;Nanjing University;University of Queensland;Shanghai Jiaotong University;Zhejiang University;National University of Singapore;Zhejiang University;Shanghai Jiaotong University;;National University of Singapore
Oral
main
Learning from Demonstration;model-based robot learning;differentiable physics-based simulation and rendering
https://openreview.net/forum?id=DYPOvNot5F
19
Diff-LfD: Contact-aware Model-based Learning from Visual Demonstration for Robotic Manipulation via Differentiable Physics-based Simulation and Rendering Learning from Demonstration (LfD) is an efficient technique for robots to acquire new skills through expert observation, significantly mitigating the need for laborio...
[ -0.05158630385994911, -0.043922051787376404, -0.02441137470304966, -0.0036225561052560806, -0.007982059381902218, -0.0013242019340395927, 0.009064449928700924, -0.0016903725918382406, -0.00298233306966722, 0.019952844828367233, -0.050149254500865936, -0.02699069119989872, -0.0025378619320690...
corl_2023_E2vL12SwO1
E2vL12SwO1
corl
2,023
PreCo: Enhancing Generalization in Co-Design of Modular Soft Robots via Brain-Body Pre-Training
Brain-body co-design, which involves the collaborative design of control strategies and morphologies, has emerged as a promising approach to enhance a robot's adaptability to its environment. However, the conventional co-design process often starts from scratch, lacking the utilization of prior knowledge. This can resu...
Yuxing Wang;Shuang Wu;Tiantian Zhang;Yongzhe Chang;Haobo Fu;QIANG FU;Xueqian Wang
Tsinghua University;Tsinghua University;Tsinghua University;Tencent AI Lab;Tencent AI Lab;Tsinghua University;Tencent AI Lab
Oral
main
Robot Co-design;Pre-training;Reinforcement Learning;Modular Soft Robots
https://openreview.net/forum?id=E2vL12SwO1
9
PreCo: Enhancing Generalization in Co-Design of Modular Soft Robots via Brain-Body Pre-Training Brain-body co-design, which involves the collaborative design of control strategies and morphologies, has emerged as a promising approach to enhance a robot's adaptability to its environment. However, the conventional co-des...
[ -0.07390577346086502, -0.04241225868463516, -0.03717571869492531, -0.005152978003025055, -0.00679171783849597, -0.0124414237216115, -0.008927186019718647, 0.03559732809662819, 0.01794721931219101, 0.02477143332362175, -0.021391822025179863, -0.015245387330651283, 0.0032519467640668154, 0.0...
corl_2023_ES_TOp4YJeD
ES_TOp4YJeD
corl
2,023
ADU-Depth: Attention-based Distillation with Uncertainty Modeling for Depth Estimation
Monocular depth estimation is challenging due to its inherent ambiguity and ill-posed nature, yet it is quite important to many applications. While recent works achieve limited accuracy by designing increasingly complicated networks to extract features with limited spatial geometric cues from a single RGB image, we int...
ZiZhang Wu;Zhuozheng Li;Zhi-Gang Fan;Yunzhe Wu;Xiaoquan Wang;Rui Tang;Jian Pu
Fudan University;ZongMu Technology Co.,Ltd.;;ZongmuTech;Zongmutech;;Fudan University
Poster
main
Monocular depth estimation;Camera perception;Distillation
https://openreview.net/forum?id=ES_TOp4YJeD
2
ADU-Depth: Attention-based Distillation with Uncertainty Modeling for Depth Estimation Monocular depth estimation is challenging due to its inherent ambiguity and ill-posed nature, yet it is quite important to many applications. While recent works achieve limited accuracy by designing increasingly complicated networks ...
[ -0.03353584185242653, -0.06225438043475151, -0.02864442579448223, -0.036166831851005554, -0.014627926982939243, -0.02645811252295971, 0.016147229820489883, -0.01583225280046463, 0.025290843099355698, 0.003024710575118661, -0.016443679109215736, -0.014136932790279388, -0.0012564361095428467, ...
corl_2023_EXQ0eXtX3OW
EXQ0eXtX3OW
corl
2,023
Dexterity from Touch: Self-Supervised Pre-Training of Tactile Representations with Robotic Play
Teaching dexterity to multi-fingered robots has been a longstanding challenge in robotics. Most prominent work in this area focuses on learning controllers or policies that either operate on visual observations or state estimates derived from vision. However, such methods perform poorly on fine-grained manipulation tas...
Irmak Guzey;Ben Evans;Soumith Chintala;Lerrel Pinto
New York University;Meta Facebook;New York University;New York University
Poster
main
Tactile;Dexterity;Manipulation
https://github.com/irmakguzey/tactile-dexterity
https://openreview.net/forum?id=EXQ0eXtX3OW
64
Dexterity from Touch: Self-Supervised Pre-Training of Tactile Representations with Robotic Play Teaching dexterity to multi-fingered robots has been a longstanding challenge in robotics. Most prominent work in this area focuses on learning controllers or policies that either operate on visual observations or state esti...
[ -0.019356602802872658, -0.0273247379809618, -0.03427909314632416, 0.025657210499048233, -0.010166242718696594, 0.03463912755250931, 0.01172955147922039, 0.017528004944324493, 0.01868390664458275, 0.014808795414865017, -0.012809655629098415, -0.0025841963943094015, 0.01636262983083725, -0.0...
corl_2023_EvuAJ0wD98
EvuAJ0wD98
corl
2,023
Intent-Aware Planning in Heterogeneous Traffic via Distributed Multi-Agent Reinforcement Learning
Navigating safely and efficiently in dense and heterogeneous traffic scenarios is challenging for autonomous vehicles (AVs) due to their inability to infer the behaviors or intentions of nearby drivers. In this work, we introduce a distributed multi-agent reinforcement learning (MARL) algorithm for joint trajectory and...
Xiyang Wu;Rohan Chandra;Tianrui Guan;Amrit Bedi;Dinesh Manocha
University of Maryland, College Park;University of Texas at Austin;Department of Computer Science, University of Maryland, College Park;University of Maryland, College Park;University of Maryland, College Park
Oral
main
Autonomous Driving;Multi-agent Reinforcement Learning;Representation Learning
https://openreview.net/forum?id=EvuAJ0wD98
10
Intent-Aware Planning in Heterogeneous Traffic via Distributed Multi-Agent Reinforcement Learning Navigating safely and efficiently in dense and heterogeneous traffic scenarios is challenging for autonomous vehicles (AVs) due to their inability to infer the behaviors or intentions of nearby drivers. In this work, we in...
[ -0.04912016540765762, -0.03469274193048477, 0.01703030616044998, 0.011062262579798698, -0.03673786669969559, 0.012512441724538803, 0.02680971845984459, 0.002963106147944927, 0.009337850846350193, 0.01933571882545948, -0.0342465341091156, -0.02032109722495079, -0.028557371348142624, 0.01559...
corl_2023_Eyb4e3GBuuR
Eyb4e3GBuuR
corl
2,023
Energy-based Potential Games for Joint Motion Forecasting and Control
This work uses game theory as a mathematical framework to address interaction modeling in multi-agent motion forecasting and control. Despite its interpretability, applying game theory to real-world robotics, like automated driving, faces challenges such as unknown game parameters. To tackle these, we establish a conne...
Christopher Diehl;Tobias Klosek;Martin Krueger;Nils Murzyn;Timo Osterburg;Torsten Bertram
TU Dortmund University;Technische Universität Dortmund;TU Dortmund University;;Technische Universität Dortmund;
Poster
main
Trajectory Prediction;Multi-Agent Interaction;Game-Theoretic Motion Planning;Energy-based Model;Optimal Control;Autonomous Vehicles
https://github.com/rst-tu-dortmund/diff_epo_planner
https://openreview.net/forum?id=Eyb4e3GBuuR
8
Energy-based Potential Games for Joint Motion Forecasting and Control This work uses game theory as a mathematical framework to address interaction modeling in multi-agent motion forecasting and control. Despite its interpretability, applying game theory to real-world robotics, like automated driving, faces challenges ...
[ -0.01645808108150959, -0.020148910582065582, -0.02480010688304901, 0.030938932672142982, -0.036719974130392075, -0.07490874826908112, 0.03815111145377159, 0.013313344679772854, 0.01919795572757721, 0.017371373251080513, -0.016467496752738953, 0.013699375092983246, -0.014622082933783531, -0...
corl_2023_FRKBdXhkQE0
FRKBdXhkQE0
corl
2,023
FastRLAP: A System for Learning High-Speed Driving via Deep RL and Autonomous Practicing
We present a system that enables an autonomous small-scale RC car to drive aggressively from visual observations using reinforcement learning (RL). Our system, FastRLAP, trains autonomously in the real world, without human interventions, and without requiring any simulation or expert demonstrations. Our system integrat...
Kyle Stachowicz;Dhruv Shah;Arjun Bhorkar;Ilya Kostrikov;Sergey Levine
University of California, Berkeley;UC Berkeley;University of California, Berkeley;University of California, Berkeley;Google
Poster
main
reinforcement learning;offroad driving;vision-based navigation
github.com/kylestach/fastrlap-release
https://openreview.net/forum?id=FRKBdXhkQE0
27
FastRLAP: A System for Learning High-Speed Driving via Deep RL and Autonomous Practicing We present a system that enables an autonomous small-scale RC car to drive aggressively from visual observations using reinforcement learning (RL). Our system, FastRLAP, trains autonomously in the real world, without human interven...
[ -0.0650336816906929, -0.021318018436431885, 0.02722850628197193, 0.005256309639662504, -0.04457570239901543, 0.015178794972598553, 0.023147890344262123, 0.007292042952030897, -0.012708466500043869, 0.009689176455140114, -0.027502987533807755, -0.033248789608478546, 0.0060202814638614655, -...
corl_2023_FefFLN5FvIM
FefFLN5FvIM
corl
2,023
TactileVAD: Geometric Aliasing-Aware Dynamics for High-Resolution Tactile Control
Touch-based control is a promising approach to dexterous manipulation. However, existing tactile control methods often overlook tactile geometric aliasing which can compromise control performance and reliability. This type of aliasing occurs when different contact locations yield similar tactile signatures. To address ...
Miquel Oller;Dmitry Berenson;Nima Fazeli
University of Michigan - Ann Arbor;University of Michigan;University of Michigan
Poster
main
Manipulation;tactile control;high-resolution tactile sensors
https://openreview.net/forum?id=FefFLN5FvIM
2
TactileVAD: Geometric Aliasing-Aware Dynamics for High-Resolution Tactile Control Touch-based control is a promising approach to dexterous manipulation. However, existing tactile control methods often overlook tactile geometric aliasing which can compromise control performance and reliability. This type of aliasing occ...
[ -0.014947536401450634, 0.00057522690622136, -0.03176829591393471, 0.002773991320282221, -0.003928028978407383, -0.01553052943199873, 0.029417211189866066, 0.03362239897251129, -0.0094186682254076, 0.06976791471242905, -0.0069528985768556595, -0.017805153504014015, 0.0239313505589962, -0.00...
corl_2023_Ffn8Z4Q-zU
Ffn8Z4Q-zU
corl
2,023
Gesture-Informed Robot Assistance via Foundation Models
Gestures serve as a fundamental and significant mode of non-verbal communication among humans. Deictic gestures (such as pointing towards an object), in particular, offer valuable means of efficiently expressing intent in situations where language is inaccessible, restricted, or highly specialized. As a result, it is e...
Li-Heng Lin;Yuchen Cui;Yilun Hao;Fei Xia;Dorsa Sadigh
Stanford University;Stanford University;Stanford University;Google;Stanford University
Poster
main
Planning with Gestures;Human-Robot Interaction;LLM Reasoning
https://openreview.net/forum?id=Ffn8Z4Q-zU
20
Gesture-Informed Robot Assistance via Foundation Models Gestures serve as a fundamental and significant mode of non-verbal communication among humans. Deictic gestures (such as pointing towards an object), in particular, offer valuable means of efficiently expressing intent in situations where language is inaccessible,...
[ -0.016514049842953682, -0.01674026995897293, 0.018041033297777176, -0.014317831955850124, -0.00798367615789175, -0.020755672827363014, 0.017777111381292343, -0.060212187469005585, 0.002420081291347742, 0.0009490629890933633, -0.024771075695753098, -0.0059524099342525005, -0.00318828620947897...
corl_2023_G_FEL3OkiR
G_FEL3OkiR
corl
2,023
Human-in-the-Loop Task and Motion Planning for Imitation Learning
Imitation learning from human demonstrations can teach robots complex manipulation skills, but is time-consuming and labor intensive. In contrast, Task and Motion Planning (TAMP) systems are automated and excel at solving long-horizon tasks, but they are difficult to apply to contact-rich tasks. In this paper, we prese...
Ajay Mandlekar;Caelan Reed Garrett;Danfei Xu;Dieter Fox
NVIDIA;NVIDIA;NVIDIA;Department of Computer Science
Poster
main
Imitation Learning;Task and Motion Planning;Teleoperation
https://openreview.net/forum?id=G_FEL3OkiR
21
Human-in-the-Loop Task and Motion Planning for Imitation Learning Imitation learning from human demonstrations can teach robots complex manipulation skills, but is time-consuming and labor intensive. In contrast, Task and Motion Planning (TAMP) systems are automated and excel at solving long-horizon tasks, but they are...
[ -0.043488871306180954, -0.03447699174284935, 0.0000716103968443349, 0.035729680210351944, -0.016415834426879883, 0.013218672946095467, -0.015677308663725853, 0.01461158785969019, -0.01557447575032711, 0.04012344032526016, -0.029466236010193825, -0.028419211506843567, -0.011741621419787407, ...
corl_2023_GsM2qJTAg-
GsM2qJTAg-
corl
2,023
Heteroscedastic Gaussian Processes and Random Features: Scalable Motion Primitives with Guarantees
Heteroscedastic Gaussian processes (HGPs) are kernel-based, non-parametric models that can be used to infer nonlinear functions with time-varying noise. In robotics, they can be employed for learning from demonstration as motion primitives, i.e. as a model of the trajectories to be executed by the robot. HGPs provide v...
Edoardo Caldarelli;Antoine Chatalic;Adrià Colomé;Lorenzo Rosasco;Carme Torras
Universidad Politécnica de Cataluna;;Spanish National Research Council;;Institut de Robòtica i Informàtica Industrial, CSIC-UPC
Poster
main
Gaussian process regression;random features;motion primitives
https://github.com/LCSL/rff-hgp
https://openreview.net/forum?id=GsM2qJTAg-
0
Heteroscedastic Gaussian Processes and Random Features: Scalable Motion Primitives with Guarantees Heteroscedastic Gaussian processes (HGPs) are kernel-based, non-parametric models that can be used to infer nonlinear functions with time-varying noise. In robotics, they can be employed for learning from demonstration as...
[ -0.05183598771691322, -0.027859071269631386, -0.012441377155482769, 0.019965359941124916, -0.0035470863804221153, -0.020094765350222588, 0.03706531971693039, -0.00653495779260993, -0.011045650579035282, 0.010731380432844162, -0.07660782337188721, -0.018015040084719658, 0.032369762659072876, ...
corl_2023_HCWoFkGe8L4
HCWoFkGe8L4
corl
2,023
Revisiting Depth-guided Methods for Monocular 3D Object Detection by Hierarchical Balanced Depth
Monocular 3D object detection has seen significant advancements with the incorporation of depth information. However, there remains a considerable performance gap compared to LiDAR-based methods, largely due to inaccurate depth estimation. We argue that this issue stems from the commonly used pixel-wise depth map loss,...
Yi-Rong Chen;Ching-Yu Tseng;Yi-Syuan Liou;Tsung-Han Wu;Winston H. Hsu
Department of computer science and informational engineering, National Taiwan University;National Taiwan University;National Taiwan University;National Taiwan University;National Taiwan University
Poster
main
monocular 3D object detection;autonomous driving
https://openreview.net/forum?id=HCWoFkGe8L4
1
Revisiting Depth-guided Methods for Monocular 3D Object Detection by Hierarchical Balanced Depth Monocular 3D object detection has seen significant advancements with the incorporation of depth information. However, there remains a considerable performance gap compared to LiDAR-based methods, largely due to inaccurate d...
[ -0.008975986391305923, -0.06466800719499588, -0.0722169429063797, -0.009092194959521294, 0.016873551532626152, 0.0073444112204015255, 0.004190498031675816, 0.002998193260282278, -0.016297155991196632, -0.0037140408530831337, -0.07132446020841599, -0.013377984054386616, 0.008543688803911209, ...
corl_2023_HDYMjiukjn
HDYMjiukjn
corl
2,023
RoboPianist: Dexterous Piano Playing with Deep Reinforcement Learning
Replicating human-like dexterity in robot hands represents one of the largest open problems in robotics. Reinforcement learning is a promising approach that has achieved impressive progress in the last few years; however, the class of problems it has typically addressed corresponds to a rather narrow definition of dext...
Kevin Zakka;Philipp Wu;Laura Smith;Nimrod Gileadi;Taylor Howell;Xue Bin Peng;Sumeet Singh;Yuval Tassa;Pete Florence;Andy Zeng;Pieter Abbeel
University of California, Berkeley;University of California, Berkeley;University of California, Berkeley;Google DeepMind;Stanford University;Simon Fraser University;Google Brain Robotics;Google;Covariant;Google;Google
Poster
main
high-dimensional control;bi-manual dexterity
https://github.com/google-research/robopianist
https://openreview.net/forum?id=HDYMjiukjn
47
RoboPianist: Dexterous Piano Playing with Deep Reinforcement Learning Replicating human-like dexterity in robot hands represents one of the largest open problems in robotics. Reinforcement learning is a promising approach that has achieved impressive progress in the last few years; however, the class of problems it has...
[ -0.03235847130417824, -0.008481435477733612, -0.027331724762916565, 0.008782466873526573, -0.026184940710663795, 0.0007615373469889164, 0.06070318818092346, 0.019447574391961098, 0.004434238187968731, 0.03123079612851143, -0.008882810361683369, 0.011248055845499039, -0.0137614281848073, -0...
corl_2023_HEIRj51lcS
HEIRj51lcS
corl
2,023
Polybot: Training One Policy Across Robots While Embracing Variability
Reusing large datasets is crucial to scale vision-based robotic manipulators to everyday scenarios due to the high cost of collecting robotic datasets. However, robotic platforms possess varying control schemes, camera viewpoints, kinematic configurations, and end-effector morphologies, posing significant challenges wh...
Jonathan Heewon Yang;Dorsa Sadigh;Chelsea Finn
Stanford University;Stanford University;Google
Poster
main
vision-based manipulation;multi-robot generalization
https://openreview.net/forum?id=HEIRj51lcS
27
Polybot: Training One Policy Across Robots While Embracing Variability Reusing large datasets is crucial to scale vision-based robotic manipulators to everyday scenarios due to the high cost of collecting robotic datasets. However, robotic platforms possess varying control schemes, camera viewpoints, kinematic configur...
[ -0.04285649210214615, -0.02784399688243866, -0.05590609461069107, 0.03624081611633301, -0.04140249639749527, -0.007192729506641626, 0.0019981067162007093, 0.008628549054265022, -0.004496023990213871, 0.006184021010994911, -0.022118892520666122, -0.002567209303379059, 0.02429988607764244, 0...
corl_2023_HYka22IcV6
HYka22IcV6
corl
2,023
Online Learning for Obstacle Avoidance
We approach the fundamental problem of obstacle avoidance for robotic systems via the lens of online learning. In contrast to prior work that either assumes worst-case realizations of uncertainty in the environment or a stationary stochastic model of uncertainty, we propose a method that is efficient to implement and p...
David Snyder;Meghan Booker;Nathaniel Simon;Wenhan Xia;Daniel Suo;Elad Hazan;Anirudha Majumdar
Princeton University;Princeton University;Princeton University;Princeton University;;Princeton University;Princeton University
Poster
main
Regret;Online;Learning;Convex;Optimization;Obstacle
https://openreview.net/forum?id=HYka22IcV6
3
Online Learning for Obstacle Avoidance We approach the fundamental problem of obstacle avoidance for robotic systems via the lens of online learning. In contrast to prior work that either assumes worst-case realizations of uncertainty in the environment or a stationary stochastic model of uncertainty, we propose a meth...
[ -0.02541368640959263, -0.04827285557985306, -0.016285045072436333, 0.025977181270718575, -0.020868148654699326, -0.003099229885265231, 0.028231168165802956, 0.0004889504634775221, 0.040083374828100204, 0.05135330185294151, -0.024436958134174347, -0.0011909729801118374, 0.028851013630628586, ...
corl_2023_HtJE9ly5dT
HtJE9ly5dT
corl
2,023
Generative Skill Chaining: Long-Horizon Skill Planning with Diffusion Models
Long-horizon tasks, usually characterized by complex subtask dependencies, present a significant challenge in manipulation planning. Skill chaining is a practical approach to solving unseen tasks by combining learned skill priors. However, such methods are myopic if sequenced greedily and face scalability issues with s...
Utkarsh Aashu Mishra;Shangjie Xue;Yongxin Chen;Danfei Xu
Sony R&D US Labs;Georgia Institute of Technology;Georgia Institute of Technology;NVIDIA
Poster
main
Manipulation Planning;Diffusion Models;Task and Motion Planning
https://github.com/generative-skill-chaining/gsc-code
https://openreview.net/forum?id=HtJE9ly5dT
79
Generative Skill Chaining: Long-Horizon Skill Planning with Diffusion Models Long-horizon tasks, usually characterized by complex subtask dependencies, present a significant challenge in manipulation planning. Skill chaining is a practical approach to solving unseen tasks by combining learned skill priors. However, suc...
[ -0.0727938562631607, -0.01222506444901228, -0.019385725259780884, -0.017400775104761124, 0.011464476585388184, -0.012670286931097507, -0.013273192569613457, 0.021185165271162987, -0.009382134303450584, 0.020665738731622696, -0.03632272407412529, -0.004677153658121824, -0.02081414684653282, ...
corl_2023_IM8zOC94HF
IM8zOC94HF
corl
2,023
Fighting Uncertainty with Gradients: Offline Reinforcement Learning via Diffusion Score Matching
Gradient-based methods enable efficient search capabilities in high dimensions. However, in order to apply them effectively in offline optimization paradigms such as offline Reinforcement Learning (RL) or Imitation Learning (IL), we require a more careful consideration of how uncertainty estimation interplays with firs...
H.J. Terry Suh;Glen Chou;Hongkai Dai;Lujie Yang;Abhishek Gupta;Russ Tedrake
Massachusetts Institute of Technology;Massachusetts Institute of Technology;Toyota Research Institute;Massachusetts Institute of Technology;University of Washington;Massachusetts Institute of Technology
Poster
main
Model-Based Reinforcement Learning;Offline Reinforcement Learning;Planning under Uncertainty;Diffusion;Score Matching
https://github.com/hjsuh94/score_po
https://openreview.net/forum?id=IM8zOC94HF
11
Fighting Uncertainty with Gradients: Offline Reinforcement Learning via Diffusion Score Matching Gradient-based methods enable efficient search capabilities in high dimensions. However, in order to apply them effectively in offline optimization paradigms such as offline Reinforcement Learning (RL) or Imitation Learning...
[ -0.07073187828063965, -0.0010396973229944706, -0.03304777294397354, -0.003785564796999097, -0.005429144483059645, -0.019806409254670143, 0.0385000966489315, 0.0014384218957275152, 0.01839696429669857, 0.0016354660037904978, -0.03673829138278961, 0.015040259808301926, -0.03501357510685921, ...
corl_2023_Id4b5SY1Y8
Id4b5SY1Y8
corl
2,023
PairwiseNet: Pairwise Collision Distance Learning for High-dof Robot Systems
Motion planning for robot manipulation systems operating in complex environments remains a challenging problem. It requires the evaluation of both the collision distance and its derivative. Owing to its computational complexity, recent studies have attempted to utilize data-driven approaches to learn the collision dist...
Jihwan Kim;Frank C. Park
Seoul National University;Seoul National University
Poster
main
Robot Collision;Collision Distance;Machine Learning
https://github.com/kjh6526/PairwiseNet
https://openreview.net/forum?id=Id4b5SY1Y8
5
PairwiseNet: Pairwise Collision Distance Learning for High-dof Robot Systems Motion planning for robot manipulation systems operating in complex environments remains a challenging problem. It requires the evaluation of both the collision distance and its derivative. Owing to its computational complexity, recent studies...
[ -0.05343443900346756, -0.013804502785205841, -0.02851899340748787, -0.0024501399602741003, -0.011120040901005268, 0.012703418731689453, 0.002973382594063878, -0.00995525810867548, 0.005432622507214546, 0.02123909629881382, -0.04757412150502205, -0.017489951103925705, 0.003919769078493118, ...
corl_2023_IeKC9khX5jD
IeKC9khX5jD
corl
2,023
Affordance-Driven Next-Best-View Planning for Robotic Grasping
Grasping occluded objects in cluttered environments is an essential component in complex robotic manipulation tasks. In this paper, we introduce an AffordanCE-driven Next-Best-View planning policy (ACE-NBV) that tries to find a feasible grasp for target object via continuously observing scenes from new viewpoints. This...
Xuechao Zhang;Dong Wang;Sun Han;Weichuang Li;Bin Zhao;Zhigang Wang;Xiaoming Duan;Chongrong Fang;Xuelong Li;Jianping He
Shanghai Jiaotong University;Shanghai AI Laboratory;;Shanghai AI Laboratory;Northwest Polytechnical University Xi'an;Shanghai Jiaotong University;Shanghai Jiaotong University;Northwestern Polytechnical University;Shanghai Jiaotong University;Shanghai AI Lab
Poster
main
Grasp Synthesis;Neural SDF;Next-Best-View Planning
https://openreview.net/forum?id=IeKC9khX5jD
14
Affordance-Driven Next-Best-View Planning for Robotic Grasping Grasping occluded objects in cluttered environments is an essential component in complex robotic manipulation tasks. In this paper, we introduce an AffordanCE-driven Next-Best-View planning policy (ACE-NBV) that tries to find a feasible grasp for target obj...
[ -0.04104211553931236, -0.05908874049782753, -0.014381485059857368, 0.03624208644032478, -0.0008872148464433849, 0.005502359941601753, 0.050084032118320465, 0.0062744575552642345, 0.026251327246427536, 0.05804687365889549, -0.01889314129948616, 0.0025232715997844934, 0.01738615520298481, 0....
corl_2023_JdpleC92J4
JdpleC92J4
corl
2,023
AR2-D2: Training a Robot Without a Robot
Diligently gathered human demonstrations serve as the unsung heroes empowering the progression of robot learning. Today, demonstrations are collected by training people to use specialized controllers, which (tele-)operate robots to manipulate a small number of objects. By contrast, we introduce AR2-D2: a system for col...
Jiafei Duan;Yi Ru Wang;Mohit Shridhar;Dieter Fox;Ranjay Krishna
University of Washington;University of Washington;Department of Computer Science, University of Washington;Department of Computer Science;University of Washington
Poster
main
Demonstration collection;Imitation learning;Augmented reality;Manipulating personalized objects;Dataset collection;Behavior Cloning
https://github.com/jiafei1224/AR2-D2_Utils
https://openreview.net/forum?id=JdpleC92J4
30
AR2-D2: Training a Robot Without a Robot Diligently gathered human demonstrations serve as the unsung heroes empowering the progression of robot learning. Today, demonstrations are collected by training people to use specialized controllers, which (tele-)operate robots to manipulate a small number of objects. By contra...
[ -0.036225274205207825, -0.007660558447241783, -0.03780990093946457, 0.01949457824230194, -0.04090544581413269, -0.01755985990166664, 0.031821489334106445, -0.0014233989641070366, -0.03053167834877968, 0.01715449057519436, -0.0372571237385273, -0.03272435814142227, -0.00037657885695807636, ...
corl_2023_JkFeyEC6VXV
JkFeyEC6VXV
corl
2,023
Finetuning Offline World Models in the Real World
Reinforcement Learning (RL) is notoriously data-inefficient, which makes training on a real robot difficult. While model-based RL algorithms (world models) improve data-efficiency to some extent, they still require hours or days of interaction to learn skills. Recently, offline RL has been proposed as a framework for t...
Yunhai Feng;Nicklas Hansen;Ziyan Xiong;Chandramouli Rajagopalan;Xiaolong Wang
;University of California, San Diego;Tsinghua University;Cerenaut AI;University of California, San Diego
Oral
main
Model-Based Reinforcement Learning;Real-World Robotics
https://openreview.net/forum?id=JkFeyEC6VXV
23
Finetuning Offline World Models in the Real World Reinforcement Learning (RL) is notoriously data-inefficient, which makes training on a real robot difficult. While model-based RL algorithms (world models) improve data-efficiency to some extent, they still require hours or days of interaction to learn skills. Recently,...
[ -0.04274817928671837, -0.025324836373329163, -0.016540341079235077, 0.012971923686563969, -0.026162322610616684, -0.01224367506802082, 0.020117860287427902, 0.031059792265295982, -0.012780758552253246, 0.026053084060549736, -0.058623991906642914, -0.016777021810412407, -0.01621262915432453, ...
corl_2023_K8cGp8rVLIP
K8cGp8rVLIP
corl
2,023
Neural Field Dynamics Model for Granular Object Piles Manipulation
We present a learning-based dynamics model for granular material manipulation. Drawing inspiration from computer graphics' Eulerian approach, our method adopts a fully convolutional neural network that operates on a density field-based representation of object piles, allowing it to exploit the spatial locality of inter...
Shangjie Xue;Shuo Cheng;Pujith Kachana;Danfei Xu
Georgia Institute of Technology;Georgia Institute of Technology;;NVIDIA
Poster
main
Deformable Object Manipulation;Manipulation Planning
https://sites.google.com/view/nfd-corl23/
https://openreview.net/forum?id=K8cGp8rVLIP
8
Neural Field Dynamics Model for Granular Object Piles Manipulation We present a learning-based dynamics model for granular material manipulation. Drawing inspiration from computer graphics' Eulerian approach, our method adopts a fully convolutional neural network that operates on a density field-based representation of...
[ -0.04090631753206253, -0.05280235782265663, -0.013428657315671444, 0.002675240393728018, -0.012753576040267944, -0.05280235782265663, 0.0012760417303070426, -0.018719840794801712, 0.017351431772112846, 0.03368111699819565, -0.039884570986032486, 0.00726625369861722, -0.00009222509106621146, ...
corl_2023_MF_cS7TCYk
MF_cS7TCYk
corl
2,023
Multi-Predictor Fusion: Combining Learning-based and Rule-based Trajectory Predictors
Trajectory prediction modules are key enablers for safe and efficient planning of autonomous vehicles (AVs), particularly in highly interactive traffic scenarios. Recently, learning-based trajectory predictors have experienced considerable success in providing state-of-the-art performance due to their ability to learn ...
Sushant Veer;Apoorva Sharma;Marco Pavone
NVIDIA;NVIDIA;Stanford University
Poster
main
trajectory prediction;rule-based planning
https://openreview.net/forum?id=MF_cS7TCYk
6
Multi-Predictor Fusion: Combining Learning-based and Rule-based Trajectory Predictors Trajectory prediction modules are key enablers for safe and efficient planning of autonomous vehicles (AVs), particularly in highly interactive traffic scenarios. Recently, learning-based trajectory predictors have experienced conside...
[ -0.03908007964491844, -0.03967445343732834, -0.030443085357546806, 0.016809634864330292, -0.018193412572145462, -0.0005467775044962764, -0.009649287909269333, -0.009714297950267792, 0.03274628520011902, 0.034770868718624115, -0.0677957683801651, -0.02162034809589386, 0.00014046725118532777, ...
corl_2023_MnANx01rV2w
MnANx01rV2w
corl
2,023
CAJun: Continuous Adaptive Jumping using a Learned Centroidal Controller
We present CAJun, a novel hierarchical learning and control framework that enables legged robots to jump continuously with adaptive jumping distances. CAJun consists of a high-level centroidal policy and a low-level leg controller. In particular, we use reinforcement learning (RL) to train the centroidal policy, which ...
Yuxiang Yang;Guanya Shi;Xiangyun Meng;Wenhao Yu;Tingnan Zhang;Jie Tan;Byron Boots
Google;University of Washington;University of Washington;Google;Google;Google;
Poster
main
Jumping;Legged Locomotion;Reinforcement Learning
https://github.com/yxyang/cajun/
https://openreview.net/forum?id=MnANx01rV2w
30
CAJun: Continuous Adaptive Jumping using a Learned Centroidal Controller We present CAJun, a novel hierarchical learning and control framework that enables legged robots to jump continuously with adaptive jumping distances. CAJun consists of a high-level centroidal policy and a low-level leg controller. In particular, ...
[ -0.0773213729262352, -0.026310745626688004, -0.028717780485749245, 0.016904793679714203, -0.023885194212198257, -0.02884739078581333, -0.06039806827902794, 0.013414592482149601, -0.029347313567996025, 0.02055237628519535, -0.03532787039875984, 0.012433262541890144, -0.013942288234829903, 0...
corl_2023_N3VbFUpwaa
N3VbFUpwaa
corl
2,023
Generalization of Heterogeneous Multi-Robot Policies via Awareness and Communication of Capabilities
Recent advances in multi-agent reinforcement learning (MARL) are enabling impressive coordination in heterogeneous multi-robot teams. However, existing approaches often overlook the challenge of generalizing learned policies to teams of new compositions, sizes, and robots. While such generalization might not be importa...
Pierce Howell;Max Rudolph;Reza Joseph Torbati;Kevin Fu;Harish Ravichandar
Georgia Institute of Technology;University of Texas at Austin;Georgia Institute of Technology;Georgia Institute of Technology;Georgia Institute of Technology
Poster
main
Heterogeneity;Multi-Robot Teaming;Generalization
https://openreview.net/forum?id=N3VbFUpwaa
6
Generalization of Heterogeneous Multi-Robot Policies via Awareness and Communication of Capabilities Recent advances in multi-agent reinforcement learning (MARL) are enabling impressive coordination in heterogeneous multi-robot teams. However, existing approaches often overlook the challenge of generalizing learned pol...
[ -0.026965558528900146, -0.005050297826528549, 0.011350532993674278, 0.013776881620287895, -0.016837390139698982, -0.03804956004023552, -0.014852195046842098, -0.005514428950846195, -0.015339303761720657, -0.019925469532608986, -0.045659471303224564, -0.015219824388623238, -0.0083130020648241...
corl_2023_Nii0_rRJwN
Nii0_rRJwN
corl
2,023
CALAMARI: Contact-Aware and Language conditioned spatial Action MApping for contact-RIch manipulation
Making contact with purpose is a central part of robot manipulation and remains essential for many household tasks -- from sweeping dust into a dustpan, to wiping tables; from erasing whiteboards, to applying paint. In this work, we investigate learning language-conditioned, vision-based manipulation policies wherein t...
Youngsun Wi;Mark Van der Merwe;Pete Florence;Andy Zeng;Nima Fazeli
University of Michigan;University of Michigan - Ann Arbor;Google;University of Michigan;Google
Poster
main
Contact-rich Manipulation;Visual-language guided policies
https://openreview.net/forum?id=Nii0_rRJwN
3
CALAMARI: Contact-Aware and Language conditioned spatial Action MApping for contact-RIch manipulation Making contact with purpose is a central part of robot manipulation and remains essential for many household tasks -- from sweeping dust into a dustpan, to wiping tables; from erasing whiteboards, to applying paint. In...
[ -0.011346819810569286, -0.0021368141751736403, -0.023529330268502235, 0.008779392577707767, -0.042490217834711075, -0.015655267983675003, 0.0027159941382706165, 0.01318533904850483, -0.0039277453906834126, 0.03966744244098663, -0.02421645261347294, 0.014736008830368519, 0.01537670474499464, ...
corl_2023_PK2debCKaG
PK2debCKaG
corl
2,023
Language Conditioned Traffic Generation
Simulation forms the backbone of modern self-driving development. Simulators help develop, test, and improve driving systems without putting humans, vehicles, or their environment at risk. However, simulators face a major challenge: They rely on realistic, scalable, yet interesting content. While recent advances in ren...
Shuhan Tan;Boris Ivanovic;Xinshuo Weng;Marco Pavone;Philipp Kraehenbuehl
NVIDIA;NVIDIA;NVIDIA;Stanford University;Apple
Poster
main
Self-driving;Content generation;Large language model
https://github.com/Ariostgx/lctgen
https://openreview.net/forum?id=PK2debCKaG
63
Language Conditioned Traffic Generation Simulation forms the backbone of modern self-driving development. Simulators help develop, test, and improve driving systems without putting humans, vehicles, or their environment at risk. However, simulators face a major challenge: They rely on realistic, scalable, yet interesti...
[ -0.06465770304203033, -0.023811407387256622, 0.008480005897581577, -0.001974924234673381, -0.021939441561698914, -0.010951001197099686, -0.02706862799823284, 0.031074635684490204, 0.00381413078866899, -0.003084063995629549, -0.03427569940686226, -0.025383859872817993, 0.00009411016799276695,...
corl_2023_PalhNjBJqv
PalhNjBJqv
corl
2,023
A Data-efficient Neural ODE Framework for Optimal Control of Soft Manipulators
This paper introduces a novel approach for modeling continuous forward kinematic models of soft continuum robots by employing Augmented Neural ODE (ANODE), a cutting-edge family of deep neural network models. To the best of our knowledge, this is the first application of ANODE in modeling soft continuum robots. This fo...
Mohammadreza Kasaei;Keyhan Kouhkiloui Babarahmati;Zhibin Li;Mohsen Khadem
University of Edinburgh, University of Edinburgh;University of Edinburgh, University of Edinburgh;University College London, University of London;Edinburgh University, University of Edinburgh
Poster
main
Soft robots;Non-parametric modelling;Optimal control
https://github.com/MohammadKasaei/SoftRobotSimulator
https://openreview.net/forum?id=PalhNjBJqv
5
A Data-efficient Neural ODE Framework for Optimal Control of Soft Manipulators This paper introduces a novel approach for modeling continuous forward kinematic models of soft continuum robots by employing Augmented Neural ODE (ANODE), a cutting-edge family of deep neural network models. To the best of our knowledge, th...
[ -0.0424937941133976, -0.02976788394153118, -0.032138943672180176, 0.02908250130712986, -0.006682491861283779, -0.011846025474369526, 0.033102184534072876, 0.023136332631111145, 0.026044582948088646, 0.021191325038671494, -0.03971521183848381, -0.016310282051563263, -0.012411003932356834, 0...
corl_2023_PsV65r0itpo
PsV65r0itpo
corl
2,023
Navigation with Large Language Models: Semantic Guesswork as a Heuristic for Planning
Navigation in unfamiliar environments presents a major challenge for robots: while mapping and planning techniques can be used to build up a representation of the world, quickly discovering a path to a desired goal in unfamiliar settings with such methods often requires lengthy mapping and exploration. Humans can rapid...
Dhruv Shah;Michael Robert Equi;Błażej Osiński;Fei Xia;brian ichter;Sergey Levine
UC Berkeley;University of California, Berkeley;University of Warsaw;Google;Google;Google
Poster
main
navigation;language models;planning;semantics
https://github.com/Michael-Equi/lfg-nav
https://openreview.net/forum?id=PsV65r0itpo
118
Navigation with Large Language Models: Semantic Guesswork as a Heuristic for Planning Navigation in unfamiliar environments presents a major challenge for robots: while mapping and planning techniques can be used to build up a representation of the world, quickly discovering a path to a desired goal in unfamiliar setti...
[ -0.04678523913025856, -0.029467662796378136, -0.0010065262904390693, -0.01074245385825634, -0.003704294329509139, -0.030875293537974358, -0.016539674252271652, -0.024429822340607643, -0.012677947990596294, -0.0042854053899645805, -0.03172728046774864, -0.006426950916647911, 0.006908508948981...
corl_2023_PwqiqaaEzJ
PwqiqaaEzJ
corl
2,023
MUTEX: Learning Unified Policies from Multimodal Task Specifications
Humans use different modalities, such as speech, text, images, videos, etc., to communicate their intent and goals with teammates. For robots to become better assistants, we aim to endow them with the ability to follow instructions and understand tasks specified by their human partners. Most robotic policy learning met...
Rutav Shah;Roberto Martín-Martín;Yuke Zhu
University of Texas at Austin;University of Texas at Austin;Computer Science Department, University of Texas, Austin
Poster
main
Multimodal Learning;Task Specification;Manipulation
https://github.com/UT-Austin-RPL/MUTEX
https://openreview.net/forum?id=PwqiqaaEzJ
65
MUTEX: Learning Unified Policies from Multimodal Task Specifications Humans use different modalities, such as speech, text, images, videos, etc., to communicate their intent and goals with teammates. For robots to become better assistants, we aim to endow them with the ability to follow instructions and understand task...
[ -0.017148811370134354, -0.029632410034537315, -0.028212562203407288, 0.008717312477529049, -0.03505364805459976, 0.011985729448497295, -0.0020145245362073183, -0.01539705228060484, -0.014309116639196873, 0.008113416843116283, -0.02391613833606243, -0.04547201097011566, 0.017268668860197067, ...
corl_2023_Pwsm7d0iWJD
Pwsm7d0iWJD
corl
2,023
Learning Lyapunov-Stable Polynomial Dynamical Systems Through Imitation
Imitation learning is a paradigm to address complex motion planning problems by learning a policy to imitate an expert's behavior. However, relying solely on the expert's data might lead to unsafe actions when the robot deviates from the demonstrated trajectories. Stability guarantees have previously been provided util...
Amin Abyaneh;Hsiu-Chin Lin
McGill University;McGill University
Poster
main
Imitation learning;Safe learning;Motion planning;Dynamical system;Semidefinite programming;Robotic manipulation
https://github.com/aminabyaneh/stable-imitation-policy
https://openreview.net/forum?id=Pwsm7d0iWJD
4
Learning Lyapunov-Stable Polynomial Dynamical Systems Through Imitation Imitation learning is a paradigm to address complex motion planning problems by learning a policy to imitate an expert's behavior. However, relying solely on the expert's data might lead to unsafe actions when the robot deviates from the demonstrat...
[ -0.0485280305147171, -0.041900359094142914, -0.008560741320252419, -0.010784006677567959, -0.018441401422023773, 0.0019728553015738726, 0.02224201336503029, 0.028476517647504807, 0.0036601966712623835, 0.027409350499510765, -0.03740702196955681, 0.02435762621462345, -0.033905964344739914, ...
corl_2023_Q8BGLiWn2X
Q8BGLiWn2X
corl
2,023
PLEX: Making the Most of the Available Data for Robotic Manipulation Pretraining
A rich representation is key to general robotic manipulation, but existing approaches to representation learning require large amounts of multimodal demonstrations. In this work we propose PLEX, a transformer-based architecture that learns from a small amount of task-agnostic visuomotor trajectories and a much larger a...
Garrett Thomas;Ching-An Cheng;Ricky Loynd;Felipe Vieira Frujeri;Vibhav Vineet;Mihai Jalobeanu;Andrey Kolobov
Stanford University;Microsoft Research;Microsoft Research;;;Microsoft Research;Microsoft
Poster
main
Robot learning;Robotic manipulation;Visuomotor representations
https://openreview.net/forum?id=Q8BGLiWn2X
12
PLEX: Making the Most of the Available Data for Robotic Manipulation Pretraining A rich representation is key to general robotic manipulation, but existing approaches to representation learning require large amounts of multimodal demonstrations. In this work we propose PLEX, a transformer-based architecture that learns...
[ -0.0581665113568306, -0.0376555509865284, -0.02547774277627468, 0.047717880457639694, -0.022773606702685356, 0.005725593771785498, 0.008245774544775486, 0.02468673698604107, 0.02748285047709942, 0.01001633983105421, -0.011221243999898434, -0.007500757463276386, 0.02402449958026409, 0.02437...
corl_2023_Q9ezhChqnL
Q9ezhChqnL
corl
2,023
Towards Scalable Coverage-Based Testing of Autonomous Vehicles
To deploy autonomous vehicles(AVs) in the real world, developers must understand the conditions in which the system can operate safely. To do this in a scalable manner, AVs are often tested in simulation on parameterized scenarios. In this context, it’s important to build a testing framework that partitions the scenari...
James Tu;Simon Suo;Chris Zhang;Kelvin Wong;Raquel Urtasun
Department of Computer Science, University of Toronto;Department of Computer Science, University of Toronto;Department of Computer Science, University of Toronto;Department of Computer Science, University of Toronto;Department of Computer Science, University of Toronto
Poster
main
Self-Driving;Coverage;Testing
https://openreview.net/forum?id=Q9ezhChqnL
4
Towards Scalable Coverage-Based Testing of Autonomous Vehicles To deploy autonomous vehicles(AVs) in the real world, developers must understand the conditions in which the system can operate safely. To do this in a scalable manner, AVs are often tested in simulation on parameterized scenarios. In this context, it’s imp...
[ -0.055946897715330124, -0.004225835669785738, -0.01324126310646534, -0.029464371502399445, -0.028923911973834038, 0.01039918977767229, -0.002464682562276721, -0.011806248687207699, 0.007468593772500753, 0.03678853437304497, -0.022773854434490204, 0.016921980306506157, 0.024302050471305847, ...
corl_2023_QG_ERxtDAP-
QG_ERxtDAP-
corl
2,023
Curiosity-Driven Learning of Joint Locomotion and Manipulation Tasks
Learning complex locomotion and manipulation tasks presents significant challenges, often requiring extensive engineering of, e.g., reward functions or curricula to provide meaningful feedback to the Reinforcement Learning (RL) algorithm. This paper proposes an intrinsically motivated RL approach to reduce task-specifi...
Clemens Schwarke;Victor Klemm;Matthijs van der Boon;Marko Bjelonic;Marco Hutter
ETHZ - ETH Zurich;ETHZ - ETH Zurich;;;ETHZ - ETH Zurich
Poster
main
Curiosity;Reinforcement Learning;Wheeled-Legged Robots
https://openreview.net/forum?id=QG_ERxtDAP-
19
Curiosity-Driven Learning of Joint Locomotion and Manipulation Tasks Learning complex locomotion and manipulation tasks presents significant challenges, often requiring extensive engineering of, e.g., reward functions or curricula to provide meaningful feedback to the Reinforcement Learning (RL) algorithm. This paper p...
[ -0.0819568857550621, -0.027243075892329216, 0.009950552135705948, 0.03228949382901192, -0.0250613521784544, 0.011392386630177498, 0.00495156180113554, -0.024833694100379944, -0.012084846384823322, 0.006004480179399252, -0.03382618725299835, -0.021987969055771828, -0.005174477118998766, 0.0...
corl_2023_QNPuJZyhFE
QNPuJZyhFE
corl
2,023
Imitating Task and Motion Planning with Visuomotor Transformers
Imitation learning is a powerful tool for training robot manipulation policies, allowing them to learn from expert demonstrations without manual programming or trial-and-error. However, common methods of data collection, such as human supervision, scale poorly, as they are time-consuming and labor-intensive. In contras...
Murtaza Dalal;Ajay Mandlekar;Caelan Reed Garrett;Ankur Handa;Ruslan Salakhutdinov;Dieter Fox
Carnegie Mellon University;NVIDIA;NVIDIA;Imperial College London;Department of Computer Science;School of Computer Science, Carnegie Mellon University
Poster
main
Imitation Learning;Task and Motion Planning;Transformers
https://openreview.net/forum?id=QNPuJZyhFE
56
Imitating Task and Motion Planning with Visuomotor Transformers Imitation learning is a powerful tool for training robot manipulation policies, allowing them to learn from expert demonstrations without manual programming or trial-and-error. However, common methods of data collection, such as human supervision, scale po...
[ -0.010060452856123447, -0.028627905994653702, -0.015691718086600304, 0.020860645920038223, -0.039760980755090714, -0.02448536641895771, 0.007503729313611984, -0.004799058195203543, 0.030329307541251183, 0.044310376048088074, -0.02378261461853981, -0.015765691176056862, 0.003950669430196285, ...
corl_2023_RN00jfIV-X
RN00jfIV-X
corl
2,023
General In-hand Object Rotation with Vision and Touch
We introduce Rotateit, a system that enables fingertip-based object rotation along multiple axes by leveraging multimodal sensory inputs. Our system is trained in simulation, where it has access to ground-truth object shapes and physical properties. Then we distill it to operate on realistic yet noisy simulated visuota...
Haozhi Qi;Brent Yi;Sudharshan Suresh;Mike Lambeta;Yi Ma;Roberto Calandra;Jitendra Malik
University of California, Berkeley;University of California, Berkeley;Carnegie Mellon University;Meta;University of California, Berkeley;Meta Facebook;University of California, Berkeley
Poster
main
In-Hand Object Rotation;Tactile Sensing;Reinforcement Learning;Sim2Real;Transformer;Visuotactile Manipulation
https://openreview.net/forum?id=RN00jfIV-X
106
General In-hand Object Rotation with Vision and Touch We introduce Rotateit, a system that enables fingertip-based object rotation along multiple axes by leveraging multimodal sensory inputs. Our system is trained in simulation, where it has access to ground-truth object shapes and physical properties. Then we distill ...
[ -0.04988394305109978, 0.006956916768103838, -0.012709928676486015, 0.043855227530002594, -0.015246400609612465, -0.0030580107122659683, 0.02032853290438652, 0.015770236030220985, 0.01360137015581131, -0.018279138952493668, -0.004884546156972647, -0.005316481459885836, 0.0043997676111757755, ...
corl_2023_RQ_7yVV8vA
RQ_7yVV8vA
corl
2,023
Learning to See Physical Properties with Active Sensing Motor Policies
To plan efficient robot locomotion, we must use the information about a terrain’s physics that can be inferred from color images. To this end, we train a visual perception module that predicts terrain properties using labels from a small amount of real-world proprioceptive locomotion. To ensure label precision, we intr...
Gabriel B. Margolis;Xiang Fu;Yandong Ji;Pulkit Agrawal
Massachusetts Institute of Technology;Massachusetts Institute of Technology;Massachusetts Institute of Technology;Massachusetts Institute of Technology
Poster
main
Locomotion;Vision;Navigation;Reinforcement Learning
https://openreview.net/forum?id=RQ_7yVV8vA
16
Learning to See Physical Properties with Active Sensing Motor Policies To plan efficient robot locomotion, we must use the information about a terrain’s physics that can be inferred from color images. To this end, we train a visual perception module that predicts terrain properties using labels from a small amount of r...
[ -0.056208107620477676, -0.025906387716531754, -0.006006333511322737, 0.026818977668881416, -0.03745344653725624, -0.07777503877878189, -0.02130618877708912, -0.019834868609905243, 0.04298486188054085, 0.031661294400691986, -0.02978024072945118, -0.01472250185906887, 0.00964738242328167, 0....
corl_2023_RaNAaxZfKi8
RaNAaxZfKi8
corl
2,023
One-shot Imitation Learning via Interaction Warping
Learning robot policies from few demonstrations is crucial in open-ended applications. We propose a new method, Interaction Warping, for one-shot learning SE(3) robotic manipulation policies. We infer the 3D mesh of each object in the environment using shape warping, a technique for aligning point clouds across object ...
Ondrej Biza;Skye Thompson;Kishore Reddy Pagidi;Abhinav Kumar;Elise van der Pol;Robin Walters;Thomas Kipf;Jan-Willem van de Meent;Lawson L.S. Wong;Robert Platt
Northeastern University;Brown University;Northeastern University;Northeastern university ;Microsoft Research;Northeastern University ;Northeastern University;Northeastern University;Northeastern University;Google
Poster
main
3D manipulation;imitation learning;shape warping
https://github.com/ondrejbiza/shapewarping
https://openreview.net/forum?id=RaNAaxZfKi8
13
One-shot Imitation Learning via Interaction Warping Learning robot policies from few demonstrations is crucial in open-ended applications. We propose a new method, Interaction Warping, for one-shot learning SE(3) robotic manipulation policies. We infer the 3D mesh of each object in the environment using shape warping, ...
[ -0.02056492306292057, -0.012572604231536388, 0.0011265275534242392, 0.009809598326683044, -0.0014881290262565017, -0.023865696042776108, 0.016670754179358482, 0.032896459102630615, -0.002700421027839184, -0.006341932807117701, -0.022715987637639046, -0.026628701016306877, 0.03055995889008045...
corl_2023_Rb0nGIt_kh5
Rb0nGIt_kh5
corl
2,023
Distilled Feature Fields Enable Few-Shot Language-Guided Manipulation
Self-supervised and language-supervised image models contain rich knowledge of the world that is important for generalization. Many robotic tasks, however, require a detailed understanding of 3D geometry, which is often lacking in 2D image features. This work bridges this 2D-to-3D gap for robotic manipulation by levera...
William Shen;Ge Yang;Alan Yu;Jansen Wong;Leslie Pack Kaelbling;Phillip Isola
Massachusetts Institute of Technology;Massachusetts Institute of Technology;Massachusetts Institute of Technology;Massachusetts Institute of Technology;Massachusetts Institute of Technology;Massachusetts Institute of Technology
Oral
main
Neural Fields;Foundation Models;Scene Understanding;Robot Manipulation
https://github.com/f3rm/f3rm
https://openreview.net/forum?id=Rb0nGIt_kh5
113
Distilled Feature Fields Enable Few-Shot Language-Guided Manipulation Self-supervised and language-supervised image models contain rich knowledge of the world that is important for generalization. Many robotic tasks, however, require a detailed understanding of 3D geometry, which is often lacking in 2D image features. ...
[ -0.06098349764943123, -0.019917413592338562, 0.0022022321354597807, -0.015490314923226833, 0.004363724030554295, -0.015309248119592667, -0.00846942700445652, 0.0009319311939179897, -0.011488725431263447, 0.01130765862762928, -0.03574270382523537, -0.0028201246168464422, -0.004965773317962885...
corl_2023_RcZMI8MSyE
RcZMI8MSyE
corl
2,023
Large Language Models as General Pattern Machines
We observe that pre-trained large language models (LLMs) are capable of autoregressively completing complex token sequences—from arbitrary ones procedurally generated by probabilistic context-free grammars (PCFG), to more rich spatial patterns found in the Abstraction and Reasoning Corpus (ARC), a general AI benchmark,...
Suvir Mirchandani;Fei Xia;Pete Florence;brian ichter;Danny Driess;Montserrat Gonzalez Arenas;Kanishka Rao;Dorsa Sadigh;Andy Zeng
Google;Google;Google;Google;Technische Universität Berlin;;;Stanford University;Google
Poster
main
large language models;in-context learning;language for robotics
https://openreview.net/forum?id=RcZMI8MSyE
220
Large Language Models as General Pattern Machines We observe that pre-trained large language models (LLMs) are capable of autoregressively completing complex token sequences—from arbitrary ones procedurally generated by probabilistic context-free grammars (PCFG), to more rich spatial patterns found in the Abstraction a...
[ -0.05898446962237358, -0.010511902160942554, -0.005113276187330484, 0.007492721546441317, 0.005950914695858955, -0.023159321397542953, 0.018621345981955528, 0.003332143882289529, -0.01997445337474346, -0.018538502976298332, -0.059720855206251144, -0.013825636357069016, 0.009867564775049686, ...
corl_2023_SgTPdyehXMA
SgTPdyehXMA
corl
2,023
Language to Rewards for Robotic Skill Synthesis
Large language models (LLMs) have demonstrated exciting progress in acquiring diverse new capabilities through in-context learning, ranging from logical reasoning to code-writing. Robotics researchers have also explored using LLMs to advance the capabilities of robotic control. However, since low-level robot actions ar...
Wenhao Yu;Nimrod Gileadi;Chuyuan Fu;Sean Kirmani;Kuang-Huei Lee;Montserrat Gonzalez Arenas;Hao-Tien Lewis Chiang;Tom Erez;Leonard Hasenclever;Jan Humplik;brian ichter;Ted Xiao;Peng Xu;Andy Zeng;Tingnan Zhang;Nicolas Heess;Dorsa Sadigh;Jie Tan;Yuval Tassa;Fei Xia
Google;Google DeepMind;Google;Google X;Google;;Google Deepmind;;Google DeepMind;Google DeepMind;Google;;Google;Google;Google DeepMind;Stanford University;Google;Google;Google;Google
Oral
main
Large language model (LLM);Low-level skill learning;Legged locomotion;Dexterous manipulation
https://openreview.net/forum?id=SgTPdyehXMA
326
Language to Rewards for Robotic Skill Synthesis Large language models (LLMs) have demonstrated exciting progress in acquiring diverse new capabilities through in-context learning, ranging from logical reasoning to code-writing. Robotics researchers have also explored using LLMs to advance the capabilities of robotic co...
[ -0.06624606996774673, 0.0063111307099461555, 0.00036092501250095665, -0.029720108956098557, -0.03129211440682411, 0.013223322108387947, 0.0005392202292568982, -0.0019384187180548906, 0.019252417609095573, -0.020491525530815125, -0.02585483156144619, 0.004688268527388573, -0.01903048716485500...
corl_2023_TWgoGdubPN
TWgoGdubPN
corl
2,023
Enabling Efficient, Reliable Real-World Reinforcement Learning with Approximate Physics-Based Models
We focus on developing efficient and reliable policy optimization strategies for robot learning with real-world data. In recent years, policy gradient methods have emerged as a promising paradigm for training control policies in simulation. However, these approaches often remain too data inefficient or unreliable to ...
Tyler Westenbroek;Jacob Levy;David Fridovich-Keil
;University of Texas at Austin;University of Texas at Austin
Poster
main
Model-based Reinforcement Learning;Feedback Control;Quadrupedal Locomotion
https://github.com/CLeARoboticsLab/LearningWithSimpleModels.jl
https://openreview.net/forum?id=TWgoGdubPN
3
Enabling Efficient, Reliable Real-World Reinforcement Learning with Approximate Physics-Based Models We focus on developing efficient and reliable policy optimization strategies for robot learning with real-world data. In recent years, policy gradient methods have emerged as a promising paradigm for training control p...
[ -0.03429161757230759, -0.008254884742200375, -0.019353775307536125, 0.004011592362076044, -0.05015626177191734, -0.04968377575278282, 0.02346077375113964, 0.014392666518688202, 0.04226937144994736, 0.028676297515630722, -0.05324559658765793, -0.008100418373942375, 0.02335173822939396, 0.02...
corl_2023_TgJ8vJUVUBR
TgJ8vJUVUBR
corl
2,023
TraCo: Learning Virtual Traffic Coordinator for Cooperation with Multi-Agent Reinforcement Learning
Multi-agent reinforcement learning (MARL) has emerged as a popular technique in diverse domains due to its ability to automate system controller design and facilitate continuous intelligence learning. For instance, traffic flow is often trained with MARL to enable intelligent simulations for autonomous driving. However...
Weiwei Liu;Wei Jing;lingping Gao;Ke Guo;Gang Xu;Yong Liu
Zhejiang University;Alibaba Group;;University of Hong Kong;;Zhejiang University
Poster
main
autonomous driving;multi-agent reinforcement learning;counterfactual reasoning
https://openreview.net/forum?id=TgJ8vJUVUBR
2
TraCo: Learning Virtual Traffic Coordinator for Cooperation with Multi-Agent Reinforcement Learning Multi-agent reinforcement learning (MARL) has emerged as a popular technique in diverse domains due to its ability to automate system controller design and facilitate continuous intelligence learning. For instance, traff...
[ -0.07287409901618958, -0.03962941840291023, 0.035684823989868164, -0.014420704916119576, -0.002914872020483017, 0.01207229495048523, 0.022621793672442436, 0.011090733110904694, -0.0073112607933580875, 0.019704626873135567, -0.02898818626999855, -0.03555639460682869, -0.02282360941171646, 0...
corl_2023_Tka2U40pHz0
Tka2U40pHz0
corl
2,023
Tuning Legged Locomotion Controllers via Safe Bayesian Optimization
This paper presents a data-driven strategy to streamline the deployment of model-based controllers in legged robotic hardware platforms. Our approach leverages a model-free safe learning algorithm to automate the tuning of control gains, addressing the mismatch between the simplified model used in the control formulati...
Daniel Widmer;Dongho Kang;Bhavya Sukhija;Jonas Hübotter;Andreas Krause;Stelian Coros
ETHZ - ETH Zurich;ETHZ - ETH Zurich;ETHZ - ETH Zurich;ETH Zurich;ETH Zurich;ETHZ - ETH Zurich
Poster
main
Legged Robotics;Bayesian Optimization;Controller Tuning;Locomotion;Machine Learning;Safe Learning
https://github.com/lasgroup/gosafeopt
https://openreview.net/forum?id=Tka2U40pHz0
21
Tuning Legged Locomotion Controllers via Safe Bayesian Optimization This paper presents a data-driven strategy to streamline the deployment of model-based controllers in legged robotic hardware platforms. Our approach leverages a model-free safe learning algorithm to automate the tuning of control gains, addressing the...
[ -0.02857351489365101, -0.027593957260251045, -0.05197200924158096, 0.011560630053281784, -0.014018765650689602, -0.013066930696368217, 0.015478860586881638, 0.013270235620439053, 0.021143661811947823, 0.03297228366136551, -0.07304174453020096, 0.006307058036327362, 0.041215356439352036, -0...
corl_2023_UVARkqnsDd
UVARkqnsDd
corl
2,023
ScalableMap: Scalable Map Learning for Online Long-Range Vectorized HD Map Construction
We propose a novel end-to-end pipeline for online long-range vectorized high-definition (HD) map construction using on-board camera sensors. The vectorized representation of HD maps, employing polylines and polygons to represent map elements, is widely used by downstream tasks. However, previous schemes designed with r...
Jingyi Yu;Zizhao Zhang;Shengfu Xia;Jizhang Sang
Wuhan University;Wuhan University;Wuhan University;Wuhan University
Poster
main
Map Construction;Multi-view Perception;Long-range Perception
https://github.com/jingy1yu/ScalableMap
https://openreview.net/forum?id=UVARkqnsDd
18
ScalableMap: Scalable Map Learning for Online Long-Range Vectorized HD Map Construction We propose a novel end-to-end pipeline for online long-range vectorized high-definition (HD) map construction using on-board camera sensors. The vectorized representation of HD maps, employing polylines and polygons to represent map...
[ -0.053934819996356964, -0.021946396678686142, -0.035457488149404526, 0.027880322188138962, 0.008750258013606071, -0.005742214620113373, 0.016605863347649574, 0.017673969268798828, -0.0034873224794864655, 0.034179411828517914, -0.007262212224304676, -0.0073306807316839695, 0.05302191153168678...
corl_2023_UZpWSDA3tZJ
UZpWSDA3tZJ
corl
2,023
Towards General Single-Utensil Food Acquisition with Human-Informed Actions
Food acquisition with common general-purpose utensils is a necessary component of robot applications like in-home assistive feeding. Learning acquisition policies in this space is difficult in part because any model will need to contend with extensive state and actions spaces. Food is extremely diverse and generally di...
Ethan Kroll Gordon;Amal Nanavati;Ramya Challa;Bernie Hao Zhu;Taylor Annette Kessler Faulkner;Siddhartha Srinivasa
Department of Computer Science, University of Washington;Department of Computer Science;Oregon State University;;;University of Washington
Poster
main
Manipulation;Learning from Demonstration;Assistive Robotics
https://github.com/personalrobotics/corl23_towards_general_food_acquisition (implementation: https://github.com/personalrobotics/ada_feeding)
https://openreview.net/forum?id=UZpWSDA3tZJ
11
Towards General Single-Utensil Food Acquisition with Human-Informed Actions Food acquisition with common general-purpose utensils is a necessary component of robot applications like in-home assistive feeding. Learning acquisition policies in this space is difficult in part because any model will need to contend with ex...
[ -0.051456574350595474, -0.01726699434220791, -0.024108633399009705, 0.010974547825753689, -0.03339837118983269, -0.0029460936784744263, -0.005352302920073271, -0.015731116756796837, -0.03118298202753067, 0.018262987956404686, 0.0001352620020043105, -0.02118581160902977, -0.01396252866834402,...
corl_2023_VH6WIPF4Sj
VH6WIPF4Sj
corl
2,023
Predicting Object Interactions with Behavior Primitives: An Application in Stowing Tasks
Stowing, the task of placing objects in cluttered shelves or bins, is a common task in warehouse and manufacturing operations. However, this task is still predominantly carried out by human workers as stowing is challenging to automate due to the complex multi-object interactions and long-horizon nature of the task. Pr...
Haonan Chen;Yilong Niu;Kaiwen Hong;Shuijing Liu;Yixuan Wang;Yunzhu Li;Katherine Rose Driggs-Campbell
University of Illinois Urbana-Champaign;University of Illinois Urbana-Champaign;UIUC;University of Illinois, Urbana Champaign;University of Illinois, Urbana Champaign;Stanford University;
Oral
main
Robotic Manipulation;Model Learning;Graph-Based Neural Dynamics;Multi-Object Interactions
https://openreview.net/forum?id=VH6WIPF4Sj
11
Predicting Object Interactions with Behavior Primitives: An Application in Stowing Tasks Stowing, the task of placing objects in cluttered shelves or bins, is a common task in warehouse and manufacturing operations. However, this task is still predominantly carried out by human workers as stowing is challenging to auto...
[ -0.03561163321137428, -0.05770644545555115, -0.030542824417352676, 0.025325478985905647, 0.0031076627783477306, -0.01871560327708721, 0.0005372844752855599, -0.016914598643779755, 0.005022391676902771, 0.014287358149886131, -0.04545218124985695, -0.01765727996826172, -0.016385436058044434, ...