Study On Pedestrian Collision Avoidance Control Strategy Of Automobile Autonomous Emergency Braking

Autonomous Emergency Braking System(AEB)is an Advanced Driving Assistant System(ADAS),which can avoid collisions of vehicles and pedestrians.It also improve the active safety of vehicles.At present,the research of AEB system mainly focuses on the realization of vehicle rear-end collision avoidance.To achieve pedestrian collision avoidance,AEB system will face more technical difficulties and higher functional requirements.This paper mainly studies how to protect the safety of pedestrians,deal with the complicated pedestrian collision avoidance test conditions,solve more flexible pedestrian trajectory and uncertain factors.The main contents are as follows:Aiming at the technical difficulties of AEB Pedestrian System(AEB-P),the functional requirements of the AEB-P system are put forward.The architecture of AEB-P system is established,and the basic functions and logical relationships of each module of this system are defined.The dynamic model of an E-class SUV is established by CarSim software and Matlab/Simulink software,and the inverse dynamic model of the vehicle is deduced theoretically.To meet the braking control requirements of AEB-P system,a theoretical model of BP feedforward fuzzy neural network is constructed,and its error back propagation algorithm is deduced.Aiming at the problem that the initial values of training parameters of fuzzy neural network can only be generated randomly,which may affect the training effect and produce control errors,genetic algorithm is introduced to optimize the theoretical control model.The optimized AEB-P system control model has the characteristics of real-time adjusting braking force,self-learning and adapting to different drivers' driving habits,and reflecting skilled drivers' emergency braking operation experience.The driving information of AEB-P system is determined,and the relevant calculation and processing are carried out.Based on the theory of Time to Collision(TTC)and braking safety distance,the early warning model of AEB-P system is established,and the safety grade of driving is divided.The range of TTC value and corresponding safety grade under different speed conditions are obtained.The working area of AEB-P early warning system is defined.The upper fuzzy neural network controller of AEB-P system is designed.The BP neural network is trained with the pedestrian longitudinal collision avoidance data of experienced drivers,and the results before and after the optimization of genetic algorithm are compared and analyzed.The results show that the braking control error of the optimized theoretical control model decreases significantly,the training efficiency improves significantly,and the stability and functional safety of the AEB-P system are enhanced.Based on the PID theory,the lower controller of AEB-P system is designed,which realizes the conversion from the expected deceleration to the braking pipeline pressure of the vehicle.Based on the domestic pedestrian collision avoidance test standard,the setting parameters of relevant test scenarios are determined,the three-dimensional pedestrian model is established,and the relevant pedestrian test scenarios are built in Carsim software.CarSim and Simulink joint simulation models of AEB-P system are established and simulated to verify the correctness of the proposed strategy.