Research On Local Path Planning For Vehicle Collision Avoidance Based On Dangerous Repellant Fields

According to the frequency and complexity of traffic accidents on straight road,a local path planning algorithm for vehicle collision avoidance based on dangerousrepellant fields was presented. In the premise of non-affecting the current trafficsituation as much as possible, the desired collision-free path can be worked out bythis algorithm with traffic environment information ahead.The general architecture of collision avoidance system was designed and therequired key technology and their relations was determined. Through referencing theartificial potential field method and inspiring from the structure of the China abacusand its operation, an abacus-type road model and the mathematical model ofdangerous repellant fields were established. The closer the host vehicle gets torepellant fields, the greater the repellant and the higher risk of collision are. First ofall, the virtual repellant field on the road boundary and obstacles boundary wereadded, and then several linear extension springs with the same stiffness were used toconnect every abacus all together. By constructing a model of China abacus-likeroad with two-lane and dangerous repellant fields, bead achieved the equilibriumwhich is based on road boundary repellant, obstacle boundary repellant and linearspring tension. Thus, a non-linear equation was got, anda non-collision trajectorywhich meets the requirements of handling&stability was solved out by usingNewton iteration method through the Mathematica software.In order to complete path following, a driver model was established based on thepreview follower theory, which met the requirement of feedback correction by usingoptimal lateral acceleration. And the simulation experiment in Simulink shows thatthe driver model had a good path following effect.Finally, through simulation with several typical working conditions, the localpath planning algorithm was investigated. The results show that the path planningalgorithm can worked out a collision-free path which also meets the performancerequirements of the vehicle dynamics. The vehicle could completed the pathfollowing well and achieved the purpose of collision avoidance finally.