Get our free extension to see links to code for papers anywhere online!

Chrome logo Add to Chrome

Firefox logo Add to Firefox
Picture for Sergey Levine

Sergey Levine

ReLMM: Practical RL for Learning Mobile Manipulation Skills Using Only Onboard Sensors


Jul 28, 2021
Charles Sun, Jędrzej Orbik, Coline Devin, Brian Yang, Abhishek Gupta, Glen Berseth, Sergey Levine

Figure 1 for ReLMM: Practical RL for Learning Mobile Manipulation Skills Using Only Onboard Sensors
Figure 2 for ReLMM: Practical RL for Learning Mobile Manipulation Skills Using Only Onboard Sensors
Figure 3 for ReLMM: Practical RL for Learning Mobile Manipulation Skills Using Only Onboard Sensors
Figure 4 for ReLMM: Practical RL for Learning Mobile Manipulation Skills Using Only Onboard Sensors
* 17 pages 
  Access Paper or Ask Questions

Persistent Reinforcement Learning via Subgoal Curricula


Jul 27, 2021
Archit Sharma, Abhishek Gupta, Sergey Levine, Karol Hausman, Chelsea Finn

Figure 1 for Persistent Reinforcement Learning via Subgoal Curricula
Figure 2 for Persistent Reinforcement Learning via Subgoal Curricula
Figure 3 for Persistent Reinforcement Learning via Subgoal Curricula
Figure 4 for Persistent Reinforcement Learning via Subgoal Curricula

  Access Paper or Ask Questions

Modularity in Reinforcement Learning via Algorithmic Independence in Credit Assignment


Jul 21, 2021
Michael Chang, Sidhant Kaushik, Sergey Levine, Thomas L. Griffiths

Figure 1 for Modularity in Reinforcement Learning via Algorithmic Independence in Credit Assignment
Figure 2 for Modularity in Reinforcement Learning via Algorithmic Independence in Credit Assignment
Figure 3 for Modularity in Reinforcement Learning via Algorithmic Independence in Credit Assignment
Figure 4 for Modularity in Reinforcement Learning via Algorithmic Independence in Credit Assignment
* Long Presentation at the Thirty-eighth International Conference on Machine Learning (ICML) 2021. 21 pages, 11 figures. v2: updated acknowledgments. v3: clarified that the internal function nodes of the credit assignment mechanism are not considered O(1) 
  Access Paper or Ask Questions

Offline Meta-Reinforcement Learning with Online Self-Supervision


Jul 19, 2021
Vitchyr H. Pong, Ashvin Nair, Laura Smith, Catherine Huang, Sergey Levine

Figure 1 for Offline Meta-Reinforcement Learning with Online Self-Supervision
Figure 2 for Offline Meta-Reinforcement Learning with Online Self-Supervision
Figure 3 for Offline Meta-Reinforcement Learning with Online Self-Supervision
Figure 4 for Offline Meta-Reinforcement Learning with Online Self-Supervision
* 10 pages, 6 figures 
  Access Paper or Ask Questions

MURAL: Meta-Learning Uncertainty-Aware Rewards for Outcome-Driven Reinforcement Learning


Jul 18, 2021
Kevin Li, Abhishek Gupta, Ashwin Reddy, Vitchyr Pong, Aurick Zhou, Justin Yu, Sergey Levine

Figure 1 for MURAL: Meta-Learning Uncertainty-Aware Rewards for Outcome-Driven Reinforcement Learning
Figure 2 for MURAL: Meta-Learning Uncertainty-Aware Rewards for Outcome-Driven Reinforcement Learning
Figure 3 for MURAL: Meta-Learning Uncertainty-Aware Rewards for Outcome-Driven Reinforcement Learning
Figure 4 for MURAL: Meta-Learning Uncertainty-Aware Rewards for Outcome-Driven Reinforcement Learning
* Accepted to ICML 2021. First two authors contributed equally 
  Access Paper or Ask Questions

Conservative Objective Models for Effective Offline Model-Based Optimization


Jul 14, 2021
Brandon Trabucco, Aviral Kumar, Xinyang Geng, Sergey Levine

Figure 1 for Conservative Objective Models for Effective Offline Model-Based Optimization
Figure 2 for Conservative Objective Models for Effective Offline Model-Based Optimization
Figure 3 for Conservative Objective Models for Effective Offline Model-Based Optimization
Figure 4 for Conservative Objective Models for Effective Offline Model-Based Optimization
* ICML 2021. First two authors contributed equally. Code at: https://github.com/brandontrabucco/design-baselines/blob/c65a53fe1e6567b740f0adf60c5db9921c1f2330/design_baselines/coms_cleaned/__init__.py 
  Access Paper or Ask Questions

Why Generalization in RL is Difficult: Epistemic POMDPs and Implicit Partial Observability


Jul 13, 2021
Dibya Ghosh, Jad Rahme, Aviral Kumar, Amy Zhang, Ryan P. Adams, Sergey Levine

Figure 1 for Why Generalization in RL is Difficult: Epistemic POMDPs and Implicit Partial Observability
Figure 2 for Why Generalization in RL is Difficult: Epistemic POMDPs and Implicit Partial Observability
Figure 3 for Why Generalization in RL is Difficult: Epistemic POMDPs and Implicit Partial Observability
Figure 4 for Why Generalization in RL is Difficult: Epistemic POMDPs and Implicit Partial Observability
* First two authors contributed equally 
  Access Paper or Ask Questions

Explore and Control with Adversarial Surprise


Jul 12, 2021
Arnaud Fickinger, Natasha Jaques, Samyak Parajuli, Michael Chang, Nicholas Rhinehart, Glen Berseth, Stuart Russell, Sergey Levine

Figure 1 for Explore and Control with Adversarial Surprise
Figure 2 for Explore and Control with Adversarial Surprise
Figure 3 for Explore and Control with Adversarial Surprise
Figure 4 for Explore and Control with Adversarial Surprise

  Access Paper or Ask Questions

Multi-Robot Deep Reinforcement Learning for Mobile Navigation


Jun 24, 2021
Katie Kang, Gregory Kahn, Sergey Levine

Figure 1 for Multi-Robot Deep Reinforcement Learning for Mobile Navigation
Figure 2 for Multi-Robot Deep Reinforcement Learning for Mobile Navigation
Figure 3 for Multi-Robot Deep Reinforcement Learning for Mobile Navigation
Figure 4 for Multi-Robot Deep Reinforcement Learning for Mobile Navigation

  Access Paper or Ask Questions

Model-Based Reinforcement Learning via Latent-Space Collocation


Jun 24, 2021
Oleh Rybkin, Chuning Zhu, Anusha Nagabandi, Kostas Daniilidis, Igor Mordatch, Sergey Levine

Figure 1 for Model-Based Reinforcement Learning via Latent-Space Collocation
Figure 2 for Model-Based Reinforcement Learning via Latent-Space Collocation
Figure 3 for Model-Based Reinforcement Learning via Latent-Space Collocation
Figure 4 for Model-Based Reinforcement Learning via Latent-Space Collocation
* International Conference on Machine Learning (ICML), 2021. Videos and code at https://orybkin.github.io/latco/ 
  Access Paper or Ask Questions

FitVid: Overfitting in Pixel-Level Video Prediction


Jun 24, 2021
Mohammad Babaeizadeh, Mohammad Taghi Saffar, Suraj Nair, Sergey Levine, Chelsea Finn, Dumitru Erhan

Figure 1 for FitVid: Overfitting in Pixel-Level Video Prediction
Figure 2 for FitVid: Overfitting in Pixel-Level Video Prediction
Figure 3 for FitVid: Overfitting in Pixel-Level Video Prediction
Figure 4 for FitVid: Overfitting in Pixel-Level Video Prediction

  Access Paper or Ask Questions

Which Mutual-Information Representation Learning Objectives are Sufficient for Control?


Jun 14, 2021
Kate Rakelly, Abhishek Gupta, Carlos Florensa, Sergey Levine

Figure 1 for Which Mutual-Information Representation Learning Objectives are Sufficient for Control?
Figure 2 for Which Mutual-Information Representation Learning Objectives are Sufficient for Control?
Figure 3 for Which Mutual-Information Representation Learning Objectives are Sufficient for Control?
Figure 4 for Which Mutual-Information Representation Learning Objectives are Sufficient for Control?
* 18 pages, 11 figures 
  Access Paper or Ask Questions

What Can I Do Here? Learning New Skills by Imagining Visual Affordances


Jun 13, 2021
Alexander Khazatsky, Ashvin Nair, Daniel Jing, Sergey Levine

Figure 1 for What Can I Do Here? Learning New Skills by Imagining Visual Affordances
Figure 2 for What Can I Do Here? Learning New Skills by Imagining Visual Affordances
Figure 3 for What Can I Do Here? Learning New Skills by Imagining Visual Affordances
Figure 4 for What Can I Do Here? Learning New Skills by Imagining Visual Affordances
* 10 pages, 10 figures. Presented at ICRA 2021. Project website: https://sites.google.com/view/val-rl 
  Access Paper or Ask Questions

Reinforcement Learning as One Big Sequence Modeling Problem


Jun 03, 2021
Michael Janner, Qiyang Li, Sergey Levine

Figure 1 for Reinforcement Learning as One Big Sequence Modeling Problem
Figure 2 for Reinforcement Learning as One Big Sequence Modeling Problem
Figure 3 for Reinforcement Learning as One Big Sequence Modeling Problem
Figure 4 for Reinforcement Learning as One Big Sequence Modeling Problem

  Access Paper or Ask Questions

Variational Empowerment as Representation Learning for Goal-Based Reinforcement Learning


Jun 02, 2021
Jongwook Choi, Archit Sharma, Honglak Lee, Sergey Levine, Shixiang Shane Gu

Figure 1 for Variational Empowerment as Representation Learning for Goal-Based Reinforcement Learning
Figure 2 for Variational Empowerment as Representation Learning for Goal-Based Reinforcement Learning
Figure 3 for Variational Empowerment as Representation Learning for Goal-Based Reinforcement Learning
Figure 4 for Variational Empowerment as Representation Learning for Goal-Based Reinforcement Learning
* Accepted at International Conference on Machine Learning (ICML) 2021 
  Access Paper or Ask Questions

Actionable Models: Unsupervised Offline Reinforcement Learning of Robotic Skills


Apr 28, 2021
Yevgen Chebotar, Karol Hausman, Yao Lu, Ted Xiao, Dmitry Kalashnikov, Jake Varley, Alex Irpan, Benjamin Eysenbach, Ryan Julian, Chelsea Finn, Sergey Levine

Figure 1 for Actionable Models: Unsupervised Offline Reinforcement Learning of Robotic Skills
Figure 2 for Actionable Models: Unsupervised Offline Reinforcement Learning of Robotic Skills
Figure 3 for Actionable Models: Unsupervised Offline Reinforcement Learning of Robotic Skills
Figure 4 for Actionable Models: Unsupervised Offline Reinforcement Learning of Robotic Skills

  Access Paper or Ask Questions

MT-Opt: Continuous Multi-Task Robotic Reinforcement Learning at Scale


Apr 27, 2021
Dmitry Kalashnikov, Jacob Varley, Yevgen Chebotar, Benjamin Swanson, Rico Jonschkowski, Chelsea Finn, Sergey Levine, Karol Hausman

Figure 1 for MT-Opt: Continuous Multi-Task Robotic Reinforcement Learning at Scale
Figure 2 for MT-Opt: Continuous Multi-Task Robotic Reinforcement Learning at Scale
Figure 3 for MT-Opt: Continuous Multi-Task Robotic Reinforcement Learning at Scale
Figure 4 for MT-Opt: Continuous Multi-Task Robotic Reinforcement Learning at Scale

  Access Paper or Ask Questions

DisCo RL: Distribution-Conditioned Reinforcement Learning for General-Purpose Policies


Apr 23, 2021
Soroush Nasiriany, Vitchyr H. Pong, Ashvin Nair, Alexander Khazatsky, Glen Berseth, Sergey Levine

Figure 1 for DisCo RL: Distribution-Conditioned Reinforcement Learning for General-Purpose Policies
Figure 2 for DisCo RL: Distribution-Conditioned Reinforcement Learning for General-Purpose Policies
Figure 3 for DisCo RL: Distribution-Conditioned Reinforcement Learning for General-Purpose Policies
Figure 4 for DisCo RL: Distribution-Conditioned Reinforcement Learning for General-Purpose Policies
* ICRA 2021 
  Access Paper or Ask Questions

Reset-Free Reinforcement Learning via Multi-Task Learning: Learning Dexterous Manipulation Behaviors without Human Intervention


Apr 22, 2021
Abhishek Gupta, Justin Yu, Tony Z. Zhao, Vikash Kumar, Aaron Rovinsky, Kelvin Xu, Thomas Devlin, Sergey Levine

Figure 1 for Reset-Free Reinforcement Learning via Multi-Task Learning: Learning Dexterous Manipulation Behaviors without Human Intervention
Figure 2 for Reset-Free Reinforcement Learning via Multi-Task Learning: Learning Dexterous Manipulation Behaviors without Human Intervention
Figure 3 for Reset-Free Reinforcement Learning via Multi-Task Learning: Learning Dexterous Manipulation Behaviors without Human Intervention
Figure 4 for Reset-Free Reinforcement Learning via Multi-Task Learning: Learning Dexterous Manipulation Behaviors without Human Intervention
* Published at ICRA 2021. First four authors contributed equally 
  Access Paper or Ask Questions

Contingencies from Observations: Tractable Contingency Planning with Learned Behavior Models


Apr 21, 2021
Nicholas Rhinehart, Jeff He, Charles Packer, Matthew A. Wright, Rowan McAllister, Joseph E. Gonzalez, Sergey Levine

Figure 1 for Contingencies from Observations: Tractable Contingency Planning with Learned Behavior Models
Figure 2 for Contingencies from Observations: Tractable Contingency Planning with Learned Behavior Models
Figure 3 for Contingencies from Observations: Tractable Contingency Planning with Learned Behavior Models
Figure 4 for Contingencies from Observations: Tractable Contingency Planning with Learned Behavior Models
* To be published at ICRA 2021. Project page: https://sites.google.com/view/contingency-planning 
  Access Paper or Ask Questions

Outcome-Driven Reinforcement Learning via Variational Inference


Apr 20, 2021
Tim G. J. Rudner, Vitchyr H. Pong, Rowan McAllister, Yarin Gal, Sergey Levine

Figure 1 for Outcome-Driven Reinforcement Learning via Variational Inference
Figure 2 for Outcome-Driven Reinforcement Learning via Variational Inference
Figure 3 for Outcome-Driven Reinforcement Learning via Variational Inference
Figure 4 for Outcome-Driven Reinforcement Learning via Variational Inference

  Access Paper or Ask Questions

RECON: Rapid Exploration for Open-World Navigation with Latent Goal Models


Apr 14, 2021
Dhruv Shah, Benjamin Eysenbach, Nicholas Rhinehart, Sergey Levine

Figure 1 for RECON: Rapid Exploration for Open-World Navigation with Latent Goal Models
Figure 2 for RECON: Rapid Exploration for Open-World Navigation with Latent Goal Models
Figure 3 for RECON: Rapid Exploration for Open-World Navigation with Latent Goal Models
Figure 4 for RECON: Rapid Exploration for Open-World Navigation with Latent Goal Models

  Access Paper or Ask Questions