Self-Supervised Mapping for Road Shape Estimation Using Laser Remission in Urban Environments

摘要

This paper describes the novel road surface analysis estimating road shape using laser scanner reflectivity in structured outdoor environments. The proposed approach can estimate road shape where a robot can drive safely in complex scenes including structures, curbs or low vegetation and so on. Road shapes are estimated robustly by using information of remission value as reflectivity of a laser, which much less depends on brightness of color or ambient lighting than passive camera. Our proposal is applicable to structured outdoor environments using road surface remission value distributions with self-supervised learning. This article shows that the method is successfully verified with road shape estimation at both the testing course of the 2009 Real World Robot Challenge, which is known as "Tsukuba Challenge" including low vegetation and our university campus.