基于LSTM深度神经网络的智能网联汽车轨迹预测

Intelligent vehicle trajectory prediction based on LSTM deep neural network

基于LSTM深度神经网络,通过实测数据分析验证了所提方法的鲁棒性和稳定性.具体选取了车辆减速测试、振荡测试、高速测试和低速测试四种实测场景,结果均显示本文所采用的轨迹预测方法可在多种复杂场景下满足工程预测的精度需求.研究结果可对智能网联汽车的车辆控制、轨迹优化以及交通管控提供理论指导.

The development of deep learning technology has brought new opportunities to improve the control and prediction accuracy of intelligent networked vehicles.Based on the LSTM deep neural network,this paper verifies the robustness and stability of the proposed method through the analysis of measured data.Specifically,four actual test scenarios are selected:vehicle deceleration test,oscillation test,high-speed test,and low-speed test.The results show that the trajectory prediction method used in this paper can meet the accuracy requirements of engineering prediction in a variety of complex scenarios.The research results can provide theoretical guidance for vehicle control,trajectory optimization and traffic control of intelligent networked vehicles.