Reinforcement Learning in RoboCup KeepAway with Partial Observability.
Sam DevlinMarek GrzesDaniel KudenkoPublished in: IAT (2009)
Keyphrases
- partial observability
- reinforcement learning
- robocup soccer
- partially observable
- temporal difference
- function approximation
- symbolic model checking
- multi agent
- markov decision process
- state space
- reinforcement learning algorithms
- partially observable markov decision processes
- fully observable
- markov decision processes
- model free
- learning algorithm
- learning agent
- planning under partial observability
- optimal policy
- belief space
- hidden state
- robotic soccer
- optimal control
- planning problems
- supervised learning
- belief state
- policy iteration
- action selection
- finite state