Approximate Soft Policy Iteration Based Reinforcement Learning for Differential Games with Two Pursuers versus One Evader.
Fan JiangXian GuoXuebo ZhangZhichao ZhangDazhi DongPublished in: ICARM (2020)
Keyphrases
- policy iteration
- policy evaluation
- reinforcement learning
- markov decision processes
- approximate policy iteration
- model free
- temporal difference
- factored mdps
- pursuit evasion
- optimal policy
- least squares
- sample path
- stochastic approximation
- markov decision process
- fixed point
- function approximation
- state space
- approximate dynamic programming
- average reward
- stochastic games
- finite state
- optimal control
- dynamic programming
- actor critic
- monte carlo
- temporal difference learning
- infinite horizon
- reinforcement learning algorithms
- learning algorithm
- game theory
- policy search
- evaluation function
- convergence rate
- sufficient conditions
- linear programming
- semi parametric
- nash equilibrium
- multi agent
- machine learning