Open Problem: Approximate Planning of POMDPs in the class of Memoryless Policies.
Kamyar AzizzadenesheliAlessandro LazaricAnimashree AnandkumarPublished in: CoRR (2016)
Keyphrases
- partially observable markov decision processes
- point based value iteration
- optimal policy
- predictive state representations
- stochastic domains
- markov decision processes
- reinforcement learning
- macro actions
- finite state
- planning problems
- markov decision problems
- belief space
- partially observable
- domain independent
- dynamic programming
- policy search
- continuous state
- planning under uncertainty
- multi agent
- belief state
- decision problems
- dynamical systems
- state space
- decision theoretic
- motion planning
- partially observable markov decision process
- heuristic search
- optimal solution