DrS: Learning Reusable Dense Rewards for Multi-Stage Tasks
This addresses the challenge of reward engineering in RL for robotics, offering a data-driven solution that is reusable across tasks, though it appears incremental as it builds on existing reward learning methods.
The paper tackles the problem of reducing human effort in reward engineering for reinforcement learning by proposing DrS, a method to learn reusable dense rewards for multi-stage tasks from sparse rewards and demonstrations, achieving improved performance and sample efficiency in robot manipulation tasks with over 1000 variants.
The success of many RL techniques heavily relies on human-engineered dense rewards, which typically demand substantial domain expertise and extensive trial and error. In our work, we propose DrS (Dense reward learning from Stages), a novel approach for learning reusable dense rewards for multi-stage tasks in a data-driven manner. By leveraging the stage structures of the task, DrS learns a high-quality dense reward from sparse rewards and demonstrations if given. The learned rewards can be \textit{reused} in unseen tasks, thus reducing the human effort for reward engineering. Extensive experiments on three physical robot manipulation task families with 1000+ task variants demonstrate that our learned rewards can be reused in unseen tasks, resulting in improved performance and sample efficiency of RL algorithms. The learned rewards even achieve comparable performance to human-engineered rewards on some tasks. See our project page (https://sites.google.com/view/iclr24drs) for more details.