Q-learning using fuzzified states and weighted actions and its application to omni-direnctional mobile robot control.
Dong-Hyun LeeIn-Won ParkJong-Hwan KimPublished in: CIRA (2009)
Keyphrases
- robot control
- state action
- reinforcement learning
- action selection
- initial state
- state transitions
- mobile robot
- state space
- motion control
- optimal policy
- evaluation function
- unstructured environments
- function approximation
- subsumption architecture
- action space
- autonomous robots
- reward function
- learning algorithm
- state transition
- markov decision processes
- reinforcement learning algorithms
- state variables
- learning rate
- average reward
- agent learns
- situation calculus
- fuzzy sets
- dynamic programming
- learning agent
- motor control
- multi agent
- machine learning