Publication
Our paper, titled Guiding Robot Exploration in Reinforcement Learning via Automated Planning, was published in the proceedings of International Conference on Automated Planning and Scheduling (ICAPS) 2021.
This work bridges the gap between symbolic automated planning and reinforcement learning, providing an effective approach to guide robot exploration in complex environments.
Research Problem
Reinforcement learning agents often struggle with exploration in large, sparse reward environments. Our research introduces automated planning techniques to provide structured guidance for exploration, significantly improving learning efficiency.
Key Innovations
- Integration of symbolic planning with reinforcement learning
- Guided exploration strategy using planning heuristics
- Theoretical analysis of convergence properties
- Experimental validation on robot navigation tasks
Results
- Faster convergence compared to standard RL methods
- Improved sample efficiency
- Better performance in complex navigation scenarios
- Successful deployment on real Segway-based robot
Authors: Yohei Hayamizu, Saeid Amiri, Kishan Chandan, Keiki Takadama, and Shiqi Zhang
Venue: International Conference on Automated Planning and Scheduling (ICAPS) 2021
| Paper | Video | Code |