Point-Based Policy Synthesis for POMDPs With Boolean and Quantitative Objectives.
Yue WangSwarat ChaudhuriLydia E. KavrakiPublished in: IEEE Robotics Autom. Lett. (2019)
Keyphrases
- point based value iteration
- partially observable markov decision processes
- belief state
- belief space
- optimal policy
- real valued
- reinforcement learning
- policy gradient
- program synthesis
- continuous state
- policy search
- partially observable
- boolean functions
- planning under uncertainty
- finite state
- qualitative and quantitative
- markov decision processes
- markov decision problems
- dynamic programming
- function approximation
- average reward
- infinite horizon
- multiple objectives
- decision problems
- policy gradient methods