Offline reinforcement learning with anderson acceleration for robotic tasks.
Guoyu ZuoShuai HuangJiangeng LiDaoxiong GongPublished in: Appl. Intell. (2022)
Keyphrases
- robotic tasks
- reinforcement learning
- motion planning
- planning under uncertainty
- markov decision processes
- dynamic environments
- robotic systems
- real time
- optimal policy
- function approximation
- multi agent
- learning algorithm
- partially observable markov decision processes
- state space
- belief space
- ai planning
- degrees of freedom
- humanoid robot
- multi robot
- optimal control
- search strategies
- heuristic search
- optimal solution
- machine learning