Equivariant Offline Reinforcement Learning
This work addresses sample efficiency for robotic manipulation tasks, but it is incremental as it adapts existing offline RL methods with equivariance for rotation-symmetric problems.
The paper tackles the problem of sample inefficiency in robotic manipulation by applying SO(2)-equivariant neural networks to offline reinforcement learning with limited demonstrations, showing that equivariant versions of CQL and IQL outperform non-equivariant ones.
Sample efficiency is critical when applying learning-based methods to robotic manipulation due to the high cost of collecting expert demonstrations and the challenges of on-robot policy learning through online Reinforcement Learning (RL). Offline RL addresses this issue by enabling policy learning from an offline dataset collected using any behavioral policy, regardless of its quality. However, recent advancements in offline RL have predominantly focused on learning from large datasets. Given that many robotic manipulation tasks can be formulated as rotation-symmetric problems, we investigate the use of $SO(2)$-equivariant neural networks for offline RL with a limited number of demonstrations. Our experimental results show that equivariant versions of Conservative Q-Learning (CQL) and Implicit Q-Learning (IQL) outperform their non-equivariant counterparts. We provide empirical evidence demonstrating how equivariance improves offline learning algorithms in the low-data regime.