Development of BP Neural Network PID Controller and Its Application on Autonomous Emergency Braking System

Abstract

Rear-end collision is one of the most common collision modes in China, which often leads to severe accident consequences. Autonomous Emergency Braking (AEB) system which can avoid or mitigate rear-end collision is one of the Advanced Driver Assistance System (ADAS) technologies. Traditional PID controller cannot effectively control the AEB system with strong nonlinear characteristics. Therefore, Back Propagation (BP) neural network PID controller is proposed in this paper. The PID parameters can be adjusted in real time based on the self-learning property and self-adapting property of BP neural network. The dynamics model is built in CarSim, and the inverse dynamics model is built in Simulink. Through the coordination control of the throttle angle and brake pressure, the host vehicle can brake automatically to avoid collisions in case of emergency. In addition, three kinds of test scenarios for the target car, stationary, slight braking, emergency braking, are setup based on complex environment in China. Finally, the simulations are conducted in these scenarios. And the simulation results indicate the feasibility and effectiveness of BP neural network PID controller in AEB system.