Corner Detection for Intelligent Vehicles Based on a Four-Layer Radar

Abstract

Intelligent vehicle is thought to be an ideal way to deal with transportation problems. Corner detection is an important yet tough problem for intelligent vehicle because more complicated pictures may be generated by visions or radars from corners than from straight roads. Four-Layer radar was selected for this task for its ability in detecting roads in more abominable natural conditions than machine vision. Radar mapping feature was firstly analyzed, and road corner trait was described based on the radar map. Road edge points as well as smoothness of road surface were the two selected characteristics to find out the road scope, and road width varying extent was the primary benchmark for corners judging. Radar data from all the four layers were combined to overcome problems possibly occur with a fast vehicle or a narrow corner. Experimental results verified a high accuracy rate of the promoted method in corner recognizing.