Map-based Multi-Policy Reinforcement Learning: Enhancing Adaptability of Robots by Deep Reinforcement Learning.
Ayaka KumeEiichi MatsumotoKuniyuki TakahashiWilson KoJethro TanPublished in: CoRR (2017)
Keyphrases
- reinforcement learning
- optimal policy
- policy search
- action selection
- function approximation
- markov decision processes
- markov decision process
- model free
- policy iteration
- learning algorithm
- robotic control
- state space
- reinforcement learning algorithms
- real robot
- action space
- approximate dynamic programming
- control problems
- robot control
- partially observable environments
- mobile robot
- state and action spaces
- actor critic
- reward function
- function approximators
- control policy
- markov decision problems
- agent learns
- continuous state spaces
- reinforcement learning problems
- partially observable
- learning capabilities
- temporal difference
- swarm intelligence
- partially observable domains
- policy evaluation
- policy gradient
- continuous state
- human users
- autonomous robots
- human robot interaction
- machine learning