Autonomous non-Line-of-Sight obstacle detection technology

Abstract

In the development of automated driving technology and advanced driver assistance systems (ADAS), autonomous non-line-of-sight (NLOS) obstacle detection without relying on infrastructure is a critical challenge. The first half of this paper introduces conventional technologies for NLOS detection using autonomous obstacle detection sensor devices. The second half describes the method which utilizes the reflection images on painted surfaces of surrounding vehicles as mirrors to estimate the location of NLOS obstacles via an on-board stereo camera, explaining its geometric principle, application examples, and principle verifications. Furthermore, this paper outlines the framework and each pipeline process for system implementation.