Decision Algorithm Development And Test Validation For Forward Collision Warning System

Forward Collision Warning system uses radar or camera sensor to detect the vehicles,pedestrians and other obstacles in front of host vehicle, based on driving states informationand sensor input information such as relative distance and relative velocity, the FCW systemdetermines whether host vehicle’s current driving status may lead to a potential rear-endcrash and how’s the degree of risk. When there is a potential risk of collision, the FCWsystem uses sounds, images or other human-computer interaction technology to sendadvance warning, causes the driver’s attention and makes him take measures to avoidcollision.State Key Laboratory of Vehicle Simulation and Control in Jilin University hasdeveloped the driver optimal preview acceleration decision algorithm, according to thealgorithm, the Forward Collision Warning system decision making algorithm with driver’spreview behavior has been researched by human-vehicle-road closed loop simulation.Based on the former simulation analysis, the following works have been carried out inthis paper: Firstly, for more accurately simulating the real behaviors of the drivers’ brakecontrol, the former single-stage warning decision algorithm has been improved, the idea ofdouble stage braking safety distance has been proposed in decision making algorithm, theweak braking safety distance and powerful braking safety distance correspond to a mildwarning and severe warning. Secondly, via the driver-in-loop simulation developed byresearch team, different types of drivers’ brake control behaviors have been researched indetail, the drivers’ weak braking and emergency braking powerful braking conditions’braking safety distance formulas have been given, and have modified the braking safetydistance formula under different road adhesion coefficient, then functional testing of thedecision-making algorithm has been done in driver-in-loop simulation. Finally, the FCWsystem decision making algorithm has been integrated in the driver assistance experimentalvehicle, multiple typical conditions of the real vehicle test validation and performance demonstration have been completed, which verified the feasibility and effectiveness ofdouble stage decision making algorithm.