Equivariant Reinforcement Learning under Partial Observability

Hai Nguyen, Andrea Baisero, David Klee, Dian Wang, Robert Platt +1 more
8/26/2024
cs.ROcs.AIcs.CV

No official code found. Best implementation: hai-h-nguyen/equi-rl-for-pomdps (8 stars, MIT, last commit 2023-11-22).

Abstract

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 group symmetries into the neural networks, our actor-critic reinforcement learning agents can reuse solutions in the past for related scenarios. Consequently, our equivariant agents outperform non-equivariant approaches significantly in terms of sample efficiency and final performance, demonstrated through experiments on a range of robotic tasks in simulation and real hardware.

View on arXivView PDF

Code Implementations(1)

hai-h-nguyen/equi-rl-for-pomdpsCommunity61%
MIT

[CoRL2023] Equivariant Reinforcement Learning under Partial Observability

81CSS, HTMLAug 5, 20232 years agoMIT
corl23equivariantpomdpreinforcement-learning

Similar Papers

Multi-Group Equivariant Augmentation for Reinforcement Learning in Robot Manipulation

85% match2025

Maximum Total Correlation Reinforcement Learning

84% match1 repo32026

E(3)-Equivariant Actor-Critic Methods for Cooperative Multi-Agent Reinforcement Learning

84% match1 repo102026

Equivariant Ensembles and Regularization for Reinforcement Learning in Map-based Path Planning

83% match1 repo1652024

Boosting Sample Efficiency and Generalization in Multi-agent Reinforcement Learning via Equivariance

83% match1 repo812026

Approximate Equivariance in Reinforcement Learning

83% match1 repo2024

Cite this paper

@article{nguyen2024equivariant,
  title  = {Equivariant Reinforcement Learning under Partial Observability},
  author = {Hai Nguyen and Andrea Baisero and David Klee and Dian Wang and Robert Platt and Christopher Amato},
  year   = {2024},
  eprint = {2408.14336},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2408.14336v1}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2408.14336.svg)](https://codeofpaper.com/paper/2408.14336)

Discussion