A surrogate-assisted controller for expensive evolutionary reinforcement learning.
Yuxing WangTiantian ZhangYongzhe ChangXueqian WangBin LiangBo YuanPublished in: Inf. Sci. (2022)
Keyphrases
- reinforcement learning
- optimal control
- function approximation
- control system
- adaptive control
- genetic algorithm
- actor critic
- control policy
- real time
- controller design
- control strategies
- state space
- robotic control
- evolutionary optimization
- markov decision processes
- neural network
- learning algorithm
- closed loop
- evolutionary methods
- transfer learning
- control strategy
- reinforcement learning methods
- control method
- pid controller
- model free
- evolutionary computation
- dc motor
- multi agent
- multi objective
- dynamic programming
- feedback control
- temporal difference
- operating conditions
- control algorithm
- optimal policy
- error prone
- reinforcement learning algorithms
- real robot
- mobile robot
- supervised learning
- computational intelligence
- dynamic model
- robust stability
- computationally expensive