基于决策树和BN的自动驾驶车辆行为决策方法

交通环境中存在着众多影响自动驾驶车辆行为决策安全的不确定性因素, 准确并及时地处理不确定性因素对自动驾驶车辆安全至关重要。因此, 建立了以人工驾驶行为分类为基础的贝叶斯网络(Bayesian network, BN)行为决策模型。利用决策树分类算法对人工驾驶行为进行分类, 利用BN建模驾驶环境并生成最优驾驶动作, 此方法既可以及时地分析人类驾驶员行为类别, 又能够充分考虑驾驶场景中的不确定性因素。利用仿真工具PreScan设计仿真实验, 仿真结果表明行为决策模型能够给出安全、合理的自动驾驶车辆行为。

Behavioral decision-making methods of autonomous vehicles based on decision tree and BN

There are many factors that affect autonomous vehicles' behavioral decision-making in the traffic environment. It is very important for the safety of autonomous vehicles to handle uncertainty factors accurately and timely. To this end, we design a behavioral decision-making model of Bayesian network (BN) based on classification of human driving behaviors. The decision tree classification algorithm is used to classify the driving behaviors of human driving vehicles, The BN is used to model the driving scene and create the best driving behavior. It not only analyzes the human driving behavioral style timely, but also takes into account the uncertain factors in the driving scene. We design the PreScan simulation experiments, the simulation results show that the behavioral decision model can provide safe and reasonable behavior of autonomous vehicles.