Wish you were here: Hindsight Goal Selection for long-horizon dexterous manipulation
This addresses sample efficiency in reinforcement learning for robotics manipulation, though it is incremental as it builds on existing hindsight relabelling methods.
The paper tackled the challenge of solving long-horizon dexterous manipulation tasks with sparse rewards by extending hindsight relabelling to use task-specific goal distributions from a small set of demonstrations, resulting in fewer required demonstrations and higher performance as task complexity increases.
Complex sequential tasks in continuous-control settings often require agents to successfully traverse a set of "narrow passages" in their state space. Solving such tasks with a sparse reward in a sample-efficient manner poses a challenge to modern reinforcement learning (RL) due to the associated long-horizon nature of the problem and the lack of sufficient positive signal during learning. Various tools have been applied to address this challenge. When available, large sets of demonstrations can guide agent exploration. Hindsight relabelling on the other hand does not require additional sources of information. However, existing strategies explore based on task-agnostic goal distributions, which can render the solution of long-horizon tasks impractical. In this work, we extend hindsight relabelling mechanisms to guide exploration along task-specific distributions implied by a small set of successful demonstrations. We evaluate the approach on four complex, single and dual arm, robotics manipulation tasks against strong suitable baselines. The method requires far fewer demonstrations to solve all tasks and achieves a significantly higher overall performance as task complexity increases. Finally, we investigate the robustness of the proposed solution with respect to the quality of input representations and the number of demonstrations.