Forward and Bidirectional Planning Based on Reinforcement Learning and Neural Networks in a Simulated Robot.
Gianluca BaldassarrePublished in: ABiALS (2003)
Keyphrases
- simulated robot
- neural network
- reinforcement learning
- real robot
- action selection
- artificial neural networks
- planning problems
- function approximators
- partially observable
- state space
- optimal policy
- back propagation
- pattern recognition
- fuzzy logic
- heuristic search
- deterministic domains
- function approximation
- blocks world
- markov decision processes
- reinforcement learning algorithms
- neural network model
- image sequences
- domain independent
- mobile robot
- learning process
- reinforcement learning problems
- partial observability
- real time
- bidirectional search
- complex domains
- learning classifier systems
- robotic systems
- recurrent neural networks
- multilayer perceptron
- degrees of freedom
- path planning
- real world