Y. He, X.Y. Zhang, Z.H. Lin

Pages: 23-32

Abstract
The automatic perception of the lateral safe distance of driverless vehicles in curve driving is the key to achieve driverless safety. The traditional perception methods have some shortcomings, such as low target vehicle location accuracy, poor tracking effect of driving route, long time of perception of horizontal safe distance, etc. In this paper, a new automatic sensing method for the safe distance of driverless vehicles in curve driving is proposed. Firstly, the dynamics model of UAV is established to obtain the driving dynamics characteristics of UAV. Then, the remote sensing technology is used to identify the target vehicle, and the particle swarm algorithm is used to estimate the driving path of the driverless vehicle. Finally, the input variables, kernel functions and model parameters of the support vector machine are selected to realize the automatic perception of the lateral safety distance of the driverless vehicle on the curve. The experimental results show that the maximum positioning accuracy of this method is 97%, the driving route tracking effect is better, and the perception time is always less than 1.5s.
Keywords: driverless vehicle; safety distance perception; dynamic model; particle swarm optimization; support vector machine