Towards Exploiting Geometry and Time for Fast Off-Distribution Adaptation in Multi-Task Robot Learning

We explore possible methods for multi-task transfer learning which seek to exploit the shared physical structure of robotics tasks. Specifically, we train policies for a base set of pre-training tasks, then experiment with adapting to new off-distribution tasks, using simple architectural approaches...

Full description

Saved in:
Bibliographic Details
Published in:arXiv.org 2021-06
Main Authors: Zentner, K R, Ryan, Julian, Puri, Ujjwal, Zhang, Yulun, Sukhatme, Gaurav
Format: Article
Language:English
Subjects:
Algorithms
Optimization
Policies
Rigid structures
Robot learning
Robotics
Training
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:We explore possible methods for multi-task transfer learning which seek to exploit the shared physical structure of robotics tasks. Specifically, we train policies for a base set of pre-training tasks, then experiment with adapting to new off-distribution tasks, using simple architectural approaches for re-using these policies as black-box priors. These approaches include learning an alignment of either the observation space or action space from a base to a target task to exploit rigid body structure, and methods for learning a time-domain switching policy across base tasks which solves the target task, to exploit temporal coherence. We find that combining low-complexity target policy classes, base policies as black-box priors, and simple optimization algorithms allows us to acquire new tasks outside the base task distribution, using small amounts of offline training data.
ISSN:2331-8422