Multi-scale Q-learning of a mobile robot in dynamic environments.
Noriko TakaseNaoyuki KubotaNorio BabaPublished in: SCIS&ISIS (2012)
Keyphrases
- dynamic environments
- mobile robot
- multiscale
- potential field
- single agent
- reinforcement learning algorithms
- path planning
- obstacle avoidance
- reinforcement learning
- function approximation
- state space
- autonomous agents
- autonomous navigation
- indoor environments
- multi robot
- cooperative
- multi agent
- autonomous robots
- real robot
- collision avoidance
- scale space
- mobile robotics
- outdoor environments
- unknown environments
- edge detection
- learning agent
- path planner
- image segmentation
- action selection
- changing environment
- optimal policy
- motion planning
- robotic systems
- model free
- simultaneous localization and mapping
- belief space
- augmented reality
- image processing
- robot control
- real environment
- reinforcement learning agents
- agent learns
- highly dynamic environments