In order to improve the real-time and computational efficiency of autonomous vehicles’ decision-making process, this paper draws on the decision-making behavior of human drivers with the motivation as the core and proposes a decision-making and planning method based on motivation and risk assessment. On the one hand, it analyzes and determines the motivations that cause the driving state to change for decision-making and planning. On the other hand, on the basis of the lateral trajectory prediction of surrounding vehicles, the longitudinal trajectory propensity prediction of different drivers is added to construct a risk assessment model that can reflect risk of the future time domain. Based on this, the motivation-based decision method is mapped into the risk assessment model, and a cost function is established to decouple the path and speed, so that the geometry and speed can be flexibly adjusted according to environmental risks. The simulation results show that the proposed method can effectively make driving behavior decisions and plan the trajectory in real time according to the current environment, which can improve the computational efficiency of the decision-making process and guarantee the safety at the same time.

中文翻译：

---

基于动机和风险评估的无人驾驶汽车决策与计划方法

为了提高自动驾驶汽车决策过程的实时性和计算效率，本文以动机为核心，以人类驾驶员的决策行为为依据，提出了基于动机的决策与计划方法。和风险评估。一方面，它分析并确定了导致驾驶状态发生变化的动机，以进行决策和计划。另一方面，基于周围车辆的横向轨迹预测，增加了不同驾驶员的纵向轨迹倾向性预测，以构建可以反映未来时域风险的风险评估模型。在此基础上，将基于动机的决策方法映射到风险评估模型中，并建立成本函数以分离路径和速度，因此可以根据环境风险灵活调整几何形状和速度。仿真结果表明，所提出的方法可以有效地根据当前环境做出驾驶行为决策并实时规划轨迹，从而可以提高决策过程的计算效率，并同时保证安全性。