Pre-training for Robots: Offline RL Enables Learning New Tasks from a Handful of Trials @rail7462
Pre-training for Robots: Offline RL Enables Learning New Tasks from a Handful of Trials  @rail7462
Uploaded October 2022 | Updated November 2026, 3 hours ago
Short video about the paper: arxiv.org/abs/2210.05178
Website: sites.google.com/view/ptr-robotlearning

Authors: {Aviral Kumar*, Anikait Singh*, Frederik Ebert*}, Yanlai Yang, Chelsea Finn, Sergey Levine
Pre-training for Robots: Offline RL Enables Learning New Tasks from a Handful of TrialsCS 182: Lecture 2, Part 4: Machine Learning BasicsCS 182: Lecture 9: Part 3: Visualization and Style TransferCS 285: Lecture 22, Meta-Learning, Part 4CS 182: Lecture 6: Part 2: Convolutional NetworksCS 185/285 (Spring 2026): Lecture 9, Off-Policy Policy GradientCS 285: Lecture 22, Part 1: Transfer Learning & Meta-LearningCS 285: Lecture 12, Part 4: Model-Based RL with PoliciesCS 182: Lecture 17: Part 1: Generative ModelsCS 185/285 (Spring 2026): Lecture 2, Supervised Learning of BehaviorsCS 182: Lecture 15: Part 2: Policy GradientsFMB: a Functional Manipulation Benchmark for Generalizable Robotic Learning
RAIL |

Pre-training for Robots: Offline RL Enables Learning New Tasks from a Handful of Trials

SHARE TO X SHARE TO REDDIT SHARE TO FACEBOOK WALLPAPER