PEAR: Primitive Enabled Adaptive Relabeling for Boosting Hierarchical Reinforcement Learning
This addresses the problem of improving training efficiency and performance in hierarchical reinforcement learning for complex robotic control, though it appears incremental as it builds on existing HRL and demonstration-based methods.
The paper tackles the difficulty of training hierarchical reinforcement learning agents due to non-stationarity by introducing PEAR, a method that uses adaptive relabeling on expert demonstrations to generate subgoal supervision and jointly optimizes with reinforcement and imitation learning, achieving up to 80% success rates in complex robotic control tasks.
Hierarchical reinforcement learning (HRL) has the potential to solve complex long horizon tasks using temporal abstraction and increased exploration. However, hierarchical agents are difficult to train due to inherent non-stationarity. We present primitive enabled adaptive relabeling (PEAR), a two-phase approach where we first perform adaptive relabeling on a few expert demonstrations to generate efficient subgoal supervision, and then jointly optimize HRL agents by employing reinforcement learning (RL) and imitation learning (IL). We perform theoretical analysis to bound the sub-optimality of our approach and derive a joint optimization framework using RL and IL. Since PEAR utilizes only a few expert demonstrations and considers minimal limiting assumptions on the task structure, it can be easily integrated with typical off-policy RL algorithms to produce a practical HRL approach. We perform extensive experiments on challenging environments and show that PEAR is able to outperform various hierarchical and non-hierarchical baselines and achieve upto $80\%$ success rates in complex sparse robotic control tasks where other baselines typically fail to show significant progress. We also perform ablations to thoroughly analyse the importance of our various design choices. Finally, we perform real world robotic experiments on complex tasks and demonstrate that PEAR consistently outperforms the baselines.