Name
Abstract | ||
|---|---|---|
Inverse Reinforcement Learning (IRL) for path planning enables robots to learn cost functions for difficult tasks from demonstration, instead of hard-coding them. However, IRL methods face practical limitations that stem from the need to repeat costly planning procedures. In this paper, we propose Rapidly Exploring Learning Trees (RLT∗), which learns the cost functions of Optimal Rapidly Exploring Random Trees (RRT∗) from demonstration, thereby making inverse learning methods applicable to more complex tasks. Our approach extends Maximum Margin Planning to work with RRT∗ cost functions. Furthermore, we propose a caching scheme that greatly reduces the computational cost of this approach. Experimental results on simulated and real-robot data from a social navigation scenario show that RLT∗ achieves better performance at lower computational cost than existing methods. We also successfully deploy control policies learned with RLT ∗ on a real telepresence robot. |
Year | DOI | Venue |
|---|---|---|
2017 | 10.1109/ICRA.2017.7989184 | ICRA |
Field | DocType | Volume |
Motion planning,Computer science,Inverse reinforcement learning,Artificial intelligence,Robot,Telerobotics,Machine learning,Trajectory,Mobile robot,Social navigation | Conference | 2017 |
Issue | Citations | PageRank |
1 | 5 | 0.46 |
References | Authors | |
11 | 3 | |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Kyriacos Shiarlis | 1 | 26 | 3.90 |
| João V. Messias | 2 | 26 | 4.77 |
| Shimon Whiteson | 3 | 1460 | 99.00 |