다중 차선 도로 차선 변경을 위한 컨볼루션 신경망 기반 충돌 감지 시스템

초록

This paper proposes a collision detection system to simultaneously predict crash situations by ego and target vehicles change lanes. There is little literature on the case research, but it is essential to predict this kind of collision for the active safety system. This paper presents a Convolution Neural Network-based collision detection system consisting of four classes to predict collision risk on the multi-lanes road. The network is formed on stacked Occupancy Grid Maps(OGMs) input data composed of point cloud data of the LiDAR and Radar sensors with in-vehicle sensor data Spatio-temporal information. The experimental results show that the collision detection system based on the proposed OGMs input data outperforms the results of other forms of OGMs, e.g., single-shot OGM and/or non-in-vehicle OGM, with an overall accuracy of 86.1% and a fall-out of about 6.1% about the collision situation from a confusion matrix.