Concurrent Modular Q-Learning with Local Rewards on Linked Multi-Component Robotic Systems.
Borja Fernández-GaunaJosé Manuel López-GuedeManuel GrañaPublished in: IWINAC (1) (2011)
Keyphrases
- robotic systems
- multi component
- reinforcement learning
- hierarchical reinforcement learning
- credit assignment
- reward function
- modular robots
- function approximation
- reinforcement learning algorithms
- vision system
- discounted reward
- state space
- markov decision processes
- control architecture
- mobile robot
- autonomous robots
- real robot
- cooperative
- optimal policy
- model free
- learning algorithm
- multi agent
- object manipulation
- multi robot
- reinforcement learning methods
- non stationary
- action selection
- imitation learning
- robotic manipulator
- stochastic approximation
- machine learning
- unstructured environments
- robot behavior
- learning agent
- dynamic programming
- search and rescue
- temporal difference
- motor control
- swarm robotics
- human teaching behavior
- reward shaping