The invention discloses a kinematics constraint-based vehicle intelligent trajectory prediction method, which comprises the following steps of: constructing a long short-term memory network model, and predicting a yaw velocity of a vehicle in a future t2 according to a lateral error, a course error and a yaw velocity of the vehicle in a past t1; based on the current real-time speed and the real-time acceleration, an attenuation acceleration model is constructed, the acceleration is attenuated to 0 in t2, and the longitudinal speed in the future t2 is iteratively solved to serve as the longitudinal speed measurement value of the vehicle in the future t2; and establishing a vehicle kinematic model, and predicting the future trajectory of the vehicle by taking the yaw velocity predicted by the long-short-term memory network and the longitudinal velocity output by the attenuation acceleration model as intermediate variables. According to the method, generalization and fault tolerance of a prediction algorithm are considered, and under the condition that the performability of the vehicle is comprehensively considered, the vehicle track prediction method combining the long-short-term memory network and the vehicle kinematic model is adopted, so that the predicted track is more accurate under the condition that the performability of the vehicle is met.
本发明公开了一种基于运动学约束的车辆智能轨迹预测方法,包括步骤:构建长短期记忆网络模型,根据车辆过去t1内的横向误差、航向误差、横摆角速度,预测未来t2内车辆的横摆角速度;基于当前实时速度与实时加速度,构建衰减加速度模型,在t2内加速度衰减到0,迭代求出未来t2内纵向速度,作为未来t2内车辆的纵向速度测量值;建立车辆运动学模型,将长短期记忆网络预测出的横摆角速度和衰减加速度模型输出的纵向速度作为中间变量,预测车辆的未来轨迹。该方法考虑预测算法的泛化性及容错性,在综合考虑车辆的可执行性的情形下,采用结合长短期记忆网络与车辆运动学模型的车辆轨迹预测方法,使预测出的轨迹在满足车辆可执行的条件下,更加准确。
Vehicle intelligent trajectory prediction method based on kinematics constraint
一种基于运动学约束的车辆智能轨迹预测方法
2023-03-07
Patent
Electronic Resource
Chinese
UTM Constraint Checking with Vehicle Trajectory
NTRS | 2016
|Attention Based Vehicle Trajectory Prediction
IEEE | 2021
|Vehicle trajectory prediction method, trajectory prediction model training method and device
European Patent Office | 2023
|