Risk-aware shielding of Partially Observable Monte Carlo Planning policies.
Giulio MazziAlberto CastelliniAlessandro FarinelliPublished in: Artif. Intell. (2023)
Keyphrases
- monte carlo
- partially observable
- markov decision problems
- state space
- partially observable markov decision processes
- reward function
- dynamical systems
- optimal policy
- markov decision processes
- reinforcement learning
- markov chain
- decision problems
- partial observability
- infinite horizon
- belief state
- planning domains
- particle filter
- monte carlo simulation
- expected utility
- planning problems
- heuristic search
- monte carlo tree search
- orders of magnitude
- dynamic programming
- optimal strategy
- markov decision process
- classical planning
- decision making
- transition probabilities
- finite state
- average cost
- bayesian networks
- decision theoretic
- utility function
- probabilistic model
- decision processes
- multi agent