基于概率统计的车辆运动轨迹预测方法

Prediction Method of Vehicle Trajectory Based on Probability Statistics

针对自主驾驶车辆预测障碍物车辆运行轨迹的问题,以高速公路上的车辆为研究对象,对其运动进行简化建模,采集大量数据作为训练样本对模型进行训练,并分别采用统计距离和马尔科夫链对其横向运动和纵向运动进行预测。仿真实验表明,该模型能够有效预测障碍物车辆3 s时间内的轨迹,可为自主驾驶车辆碰撞预警和轨迹规划提供可靠依据。

Considering the problem of autonomous vehicles predicting obstacle vehicle trajectory, the paper firstly takes vehicles on expressway as study object, and simplifies the movement and models on it. Then, it trains the model by collecting large amounts of data, and predicts its lateral and longitudinal movement with statistical distance and Markov chain respectively. The simulation experiment shows that this model can predict the trajectory of obstacle vehicle within 3s, which can provide reliable basis for collision warning and trajectory planning for autonomous vehicles.