Rover-IRL: Inverse Reinforcement Learning With Soft Value Iteration Networks for Planetary Rover Path Planning.
Max PfluegerAli-Akbar Agha-MohammadiGaurav S. SukhatmePublished in: IEEE Robotics Autom. Lett. (2019)
Keyphrases
- path planning
- inverse reinforcement learning
- bayesian nonparametric
- partially observable environments
- reward function
- mobile robot
- markov decision processes
- path planning algorithm
- preference elicitation
- dynamic environments
- autonomous navigation
- motion planning
- multi robot
- optimal path
- markov decision process
- state space
- path finding
- dynamic and uncertain environments
- policy iteration
- multiple robots
- dynamic programming
- collision free
- temporal difference
- network structure
- heuristic search
- optimal policy