Abstract: For the intelligent vehicle active collision avoidance at high speed, most of the current studies do not focus on the vehicle dynamics model, which directly limits the safety of the collision avoidance algorithm in practical application. Therefore, an improved method for intelligent vehicle high-speed active collision avoidance based on the predicted trajectory of adjacent vehicles in front and the stability of self vehicle dynamics is proposed. Data processing and feature extraction are carried out based on the Next generation simulation(NGSIM) data set. The prediction model of the Long short-term memory-recurrent neural network(LSTM-RNN) has been established and trained, prediction and validation have been carried out to obtain the multi-time domain prediction track of the preceding vehicle. According to the predicted trajectory and the dynamic stability of the preceding vehicle, a reasonable method of active lane changing and obstacle avoidance is designed. Based on Technique for order preference by similarity to ideal solution (TOPSIS) algorithm, an active collision avoidance path with safety is planned. The Prescan-Simulink-Carsim hardware-in-the-loop test platform is built to verify the proposed active collision avoidance method, and the results show that the proposed method can effectively solve the problem of collision avoidance safety in high-speed scenarios.

Key words: intelligent vehicles, collision avoidance in high-speed scene, trajectory prediction, dynamic stability, hardware- in-the-loop