Optimizing pedestrian simulation based on expert trajectory guidance and deep reinforcement learning.
Senlin MuXiao HuangMoyang WangDi ZhangDong XuXiang LiPublished in: GeoInformatica (2023)
Keyphrases
- reinforcement learning
- function approximation
- pedestrian detection
- reinforcement learning algorithms
- learning algorithm
- trajectory data
- expert knowledge
- dynamic programming
- state space
- object detection
- temporal difference learning
- human experts
- markov decision processes
- robot control
- optimal control
- domain experts
- domain specific
- reinforcement learning methods
- learning process
- data sets
- expert advice
- markov decision process
- temporal difference
- action selection
- supervised learning
- genetic algorithm