HTN-Style Planning in Relational POMDPs Using First-Order FSCs.
Felix MüllerSusanne BiundoPublished in: KI (2011)
Keyphrases
- partially observable markov decision processes
- planning problems
- decision diagrams
- predictive state representations
- finite state
- planning domains
- htn planning
- planning under uncertainty
- dynamical systems
- stochastic domains
- state space
- domain independent
- belief space
- hierarchical task network
- heuristic search
- planning systems
- relational learning
- reinforcement learning
- ai planning
- decision problems
- partial observability
- partially observable
- markov decision processes
- dynamic programming
- classical planning
- policy gradient
- optimal policy
- belief state
- relational data
- hierarchical task networks
- domain specific
- markov chain
- data model
- stochastic systems
- point based value iteration
- relational databases
- policy gradient methods
- actor critic
- action models
- average reward
- first order logic
- multi agent