Online Planning in POMDPs with Self-Improving Simulators.
Jinke HeMiguel SuauHendrik BaierMichael KaisersFrans A. OliehoekPublished in: IJCAI (2022)
Keyphrases
- partially observable markov decision processes
- belief state
- stochastic domains
- real time
- belief space
- online learning
- partially observable
- reinforcement learning
- planning problems
- markov decision problems
- markov decision processes
- linear programming
- point based value iteration
- neural network
- finite state
- planning under uncertainty
- predictive state representations
- heuristic search
- decision support
- planning systems
- motion planning
- state space
- travel planning
- optimal solution
- learning algorithm
- sequential decision making problems
- partially observable stochastic games