Discretionary Lane-Change Decision and Control via Parameterized Soft Actor-Critic for Hybrid Action Space.
Yuan LinXiao LiuZishun ZhengLiyao WangPublished in: CoRR (2024)
Keyphrases
- action space
- actor critic
- reinforcement learning
- optimal control
- action selection
- temporal difference
- control system
- state space
- control problems
- function approximation
- real valued
- learning tasks
- control strategy
- decision makers
- policy gradient
- decision making
- machine learning
- decision problems
- adaptive control
- markov decision processes
- markov chain
- single agent
- stochastic processes
- policy iteration
- bayesian networks
- gradient method