Deep Neural Network를 이용한 차량 내부 센서 기반 도로 노면 상태 판단

Abstract

In this paper, we present a method detecting a road surface condition using deep neural network based on in-vehicle sensors. Determining the road surface condition, since it is related to estimate the tire-road friction coefficient, is a core technology for future autonomous driving as well as for safety systems. To determine the road surface condition, feature vectors to be used in deep neural network are developed from parameters obtained from mathematical and/or dynamic models of the vehicle with a time-windows approach. The presented method is verified using experimental data obtained with a real vehicle on a proving ground. We observed that the performance the road surface condition detection using deep neural network is superior.