Metric State Space Reinforcement Learning for a Vision-Capable Mobile Robot.
Viktor ZhumatiyFaustino J. GomezMarcus HutterJürgen SchmidhuberPublished in: IAS (2006)
Keyphrases
- state space
- reinforcement learning
- mobile robot
- reinforcement learning algorithms
- optimal policy
- markov decision processes
- heuristic search
- computer vision
- path planning
- action space
- vision system
- dynamic programming
- markov chain
- topological map
- obstacle avoidance
- robot control
- model free
- mobile robot navigation
- real robot
- indoor environments
- markov decision process
- continuous state spaces
- partially observable
- state abstraction
- function approximation
- mobile robotics
- distance metric
- unknown environments
- multi agent
- state variables
- distance function
- autonomous robots
- dynamic environments
- particle filter
- control problems
- continuous state
- machine learning
- state and action spaces
- dynamical systems
- metric space
- learning process
- image processing
- control system
- goal state
- function approximators
- simultaneous localization and mapping
- autonomous navigation
- multi robot
- motion planning
- infinite horizon
- action selection
- temporal difference
- loop closing
- real time
- belief state
- distance measure
- markov decision problems
- motion control
- robotic systems
- optimal control
- reward function
- learning capabilities