An adaptive vehicle rear-end collision warning algorithm based on neural network

Abstract

Most of the existing algorithms of vehicle rear-end collision have poor adaptive, high false alarm and missed alarm rates. A two-level early warning model based on logic algorithm of safe distance is discussed. The influence of road conditions, driver status and vehicle performance on the warning distance of rear-end collision in the driving process is analyzed. And for different driving conditions, a warning algorithm of vehicle rear-end collision based on neural network with adaptive threshold which can adapt to different status of the three main elements, human-vehicle-road is proposed. Also the comparison of the warning distance whether using adaptive strategies for the rear-end collision algorithm through changing the real-time status of human-vehicle-road is presented. The result of the simulation shows that the algorithm proposed is self-adaptive to the warning distance and region, and the feasibility of the algorithm is verified. © 2011 Springer-Verlag.