Rule-based Shielding for Partially Observable Monte-Carlo Planning.
Giulio MazziAlberto CastelliniAlessandro FarinelliPublished in: CoRR (2021)
Keyphrases
- monte carlo
- partially observable
- state space
- decision problems
- markov chain
- dynamical systems
- markov decision processes
- partial observability
- reinforcement learning
- markov decision problems
- belief state
- infinite horizon
- belief space
- planning domains
- particle filter
- partial observations
- monte carlo simulation
- importance sampling
- monte carlo tree search
- temporal difference
- optimal strategy
- reward function
- planning problems
- partially observable markov decision processes
- optimal policy
- transition probabilities
- learning algorithm
- heuristic search
- object tracking
- probability distribution
- decision making