An online and approximate solver for POMDPs with continuous action space.
Konstantin M. SeilerHanna KurniawatiSurya P. N. SinghPublished in: ICRA (2015)
Keyphrases
- action space
- continuous state
- markov decision processes
- state space
- reinforcement learning
- continuous state spaces
- real valued
- continuous action
- state and action spaces
- point based value iteration
- markov decision problems
- dynamic programming
- control policies
- optimal policy
- belief state
- action selection
- single agent
- partially observable markov decision processes
- markov decision process
- partially observable
- reinforcement learning methods
- heuristic search
- orders of magnitude
- policy iteration
- state action
- model free
- policy search
- markov chain
- search space
- function approximation
- planning problems
- computational complexity