Offline Policy Iteration Based Reinforcement Learning Controller for Online Robotic Knee Prosthesis Parameter Tuning.
Minhan LiXiang GaoYue WenJennie SiHe Helen HuangPublished in: ICRA (2019)
Keyphrases
- parameter tuning
- policy iteration
- reinforcement learning
- markov decision processes
- real time
- policy iteration algorithm
- model free
- actor critic
- optimal control
- temporal difference
- optimal policy
- finite state
- stochastic approximation
- markov decision process
- approximate dynamic programming
- function approximation
- least squares
- policy evaluation
- state space
- fixed point
- parameter settings
- average reward
- reinforcement learning algorithms
- machine learning
- infinite horizon
- convergence rate
- dynamic programming
- closed loop
- adaptive control
- average cost
- markov decision problems
- linear programming
- fuzzy controller
- genetic algorithm
- action space
- learning algorithm
- robust stability
- step size