An Auxiliary Decision-Making Method for Autonomous Driving via Monte Carlo Tree Search
Tianxiang Ou, Yanan Lu, Xuesong Wu, Jianwen Cao · 2022
The decision-making problem in autonomous driving (AD) is one of the most challenging scenarios, for AD is a real time scenario with a huge space of actions and spaces. Besides there are multiple other agents, it is difficult to predict their actions. In this paper, we propose a fast decision-making auxiliary method based on Monte Carlo Tree Search (MCTS). First, our ego vehicle is controlled by some basic heuristic methods built by humans’ driving experiences, then we use our auxiliary method to improve the ego vehicle's decision-making ability. Besides, we use a pruning technology to reduce time cost. Experiments are carried out on the simulation platform CARLA, the results show that our auxiliary method can improve the decision-making ability of the ego vehicle, and the pruning can make the driving policy stronger.