Preceding vehicle recognition for lead vehicle of fully automated platooning trucks based on multi-sensor fusion

摘要

We have been developing an autonomous platooning system of heavy trucks aimed atenhancing its energy efficiency by shortening headway distance between trucks. In this paper,we focused on developing a forward environment perception method for fully automated leadvehicle of the autonomous platooning system. To percept the forward environment ofautonomous platooning system, it is important to robustly detect obstacles from distance.Additionally, it is also important to recognize which lane the detected obstacles exist. In thisresearch, the obstacles are detected robustly and precisely based on multi-sensor fusionscheme between LIDAR and RADAR. Moreover, the lanes, where the obstacles exist, arerecognized by using geometrical curvature of driving lane estimated from both ego-motionand lane marker position detected from cameras.