Decentralized control of Partially Observable Markov Decision Processes using belief space macro-actions.
Shayegan OmidshafieiAli-akbar Agha-mohammadiChristopher AmatoJonathan P. HowPublished in: ICRA (2015)
Keyphrases
- decentralized control
- belief space
- partially observable markov decision processes
- markov decision processes
- finite state
- planning under uncertainty
- optimal policy
- state space
- reinforcement learning
- partially observable
- dynamic programming
- continuous state
- decision theoretic planning
- infinite horizon
- belief state
- average cost
- reinforcement learning algorithms
- multi agent
- reward function
- multi agent systems
- smart grid
- collective behavior
- long run
- motion planning
- initial state
- decision problems
- dynamical systems
- dynamic environments
- decision makers
- point based value iteration