Abstract:

Aiming at the problem that the safety distance model and the collision avoidance strategy of intelligent vehicles are too simple, this paper proposes a coordinated collision avoidance strategy of vehicle braking and steering under complex working conditions.

Firstly, a vehicle system model is constructed based on the three-degree-of-freedom dynamic model combined with the magic tire formula. The advantages and disadvantages of common safety distance models are compared and analyzed, and the optimization is carried out on their basis. For the longitudinal braking mode, four longitudinal safety distance models are established based on different driving states of the vehicle in front; For the lateral lane change mode, a minimum longitudinal distance model for critical collision is established based on the driving state of the front and rear vehicles in the target lane and the safety constraints when the vehicle changes the lanes. For the collaborative collision avoidance mode, a collaborative collision avoidance safety distance model is established based on the vehicle’s high-speed lane-changing extreme conditions. The selection of different collision avoidance modes of the vehicle is based on critical distance division of the safety distance model under different working conditions, and finally a switching collision avoidance strategy that satisfies certain constraints is constructed. In the selection of lane-changing trajectory planning methods, factors such as trajectory error, trajectory curvature and vehicle dynamics constraints of different lane-changing trajectory methods are analyzed, and a quintic polynomial is selected as the reference trajectory for lane-changing. The maximum lateral acceleration of the planned trajectory and the limit value of the side slip angle of the center of mass are determined, and the planned trajectory curve is more in line with the driving trajectory during steering and collision avoidance.

Secondly, based on the design of the vehicle collision avoidance controller, the longitudinal control of the vehicle adopts fuzzy logic control. The relative speed and relative distance between the vehicle concerned and the preceding vehicle are used as the input of longitudinal collision avoidance, and the fuzzy rule output of the fuzzy controller corresponds to the expected acceleration. Vehicle lateral control adopts online LQR control algorithm combined with certain constraints. To a certain extent, the tracking effect of this controller is better than that of the model prediction control algorithm. At the same time, in the trajectory tracking process, in order to allow the vehicle to adapt to the speed of different front and rear vehicles, the speed function in the process of vehicle lane changing is designed.

Finally, a joint simulation platform of Carsim and Matlab/simulink is built to verify the effectiveness of the controller under three simulation conditions of longitudinal, lateral and cooperative collision avoidance. The longitudinal collision avoidance controller has good collision avoidance ability with different road surface adhesion coefficients and under complex driving conditions. The trajectory tracking controller has good real-time performance and good trajectory tracking effect. Under the conditions of short inter-vehicle distance and high vehicle speed, the cooperative collision avoidance strategy reduces vehicle lateral acceleration and improves vehicle lateral stability. Based on the BAIC new energy real vehicle platform, the campus road test is carried out, and the vehicle collision avoidance tests are set at different speeds and under different modes. The test results show that the effectiveness of the vehicle’s longitudinal and lateral collision avoidance controllers improves the potential of vehicle collision avoidance performance.

To sum up, this paper proposes a multi-mode vehicle collision avoidance strategy for the problems that the safety distance model is too simple and the collision avoidance strategy is too weak. Firstly, the traditional safety distance model is improved, and the decision logic for collision avoidance under different working conditions is formulated, which improves the adaptability of the vehicle to different driving conditions. Then, three simulation working conditions are set in the simulation software, and the designed collision avoidance is verified. What is more, the effectiveness of the collision controller provides an effective basis for the real vehicle test. Finally, through two groups of real vehicle tests under different modes, it is verified that the designed collision avoidance system has a good collision avoidance ability and improves the active safety performance of the vehicle.