Equivariant Reinforcement Learning under Partial Observability

Autor:	Nguyen, Hai, Baisero, Andrea, Klee, David, Wang, Dian, Platt, Robert, Amato, Christopher
Rok vydání:	2024
Předmět:	Computer Science - Robotics Computer Science - Artificial Intelligence Computer Science - Computer Vision and Pattern Recognition
Druh dokumentu:	Working Paper
Popis:	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. Comment: Conference on Robot Learning, 2023
Databáze:	arXiv
Externí odkaz:	http://arxiv.org/abs/2408.14336 Zobrazit plný text záznamu View this record from Arxiv