Using Partial-Policy Q-Learning to Plan Path for Robot Navigation in Unknown Enviroment.
Juli ZhangJunyi ZhangZhong MaZhanzhuang HePublished in: ISCID (1) (2017)
Keyphrases
- robot navigation
- optimal policy
- initially unknown
- action selection
- reinforcement learning
- state space
- autonomous mobile robot
- reward function
- continuous state
- autonomous robots
- partial plans
- goal state
- continuous state spaces
- scene understanding
- learning algorithm
- reinforcement learning algorithms
- state action
- function approximation
- map building
- dynamic programming
- markov decision processes
- multi class
- landmark recognition
- model free
- markov decision process
- action space
- control policies
- markov decision problems
- multi view
- machine learning
- agent receives