Continuously Improving Mobile Manipulation with Autonomous Real-World RL
This work addresses the challenge of enabling robots to learn and improve continuously in real-world environments, which is incremental as it builds on prior RL methods with specific enhancements for mobile manipulation.
The authors tackled the problem of autonomous real-world reinforcement learning for mobile manipulation by developing a framework that learns policies without extensive instrumentation or human supervision, achieving an average success rate of 80% across four tasks, which is a 3-4x improvement over existing approaches.
We present a fully autonomous real-world RL framework for mobile manipulation that can learn policies without extensive instrumentation or human supervision. This is enabled by 1) task-relevant autonomy, which guides exploration towards object interactions and prevents stagnation near goal states, 2) efficient policy learning by leveraging basic task knowledge in behavior priors, and 3) formulating generic rewards that combine human-interpretable semantic information with low-level, fine-grained observations. We demonstrate that our approach allows Spot robots to continually improve their performance on a set of four challenging mobile manipulation tasks, obtaining an average success rate of 80% across tasks, a 3-4 improvement over existing approaches. Videos can be found at https://continual-mobile-manip.github.io/