Drivable area detection and vehicle location based on multi-sensor information fusion 2020-01-1027
Multi-sensor information fusion framework is the eyes for unmanned driving and Advanced Driver Assistance System (ADAS) to perceive the surrounding environment. In addition to the perception of the surrounding environment, real-time vehicle location is also the key and difficult point of unmanned driving technology. The disappearance of high-precision GPS differential signal and the defect of lane line will bring much more difficult for vehicle self-locating. In this paper, a road boundary feature extraction algorithm is proposed based on multi-sensor information fusion of automotive radar and vision to realize the auxiliary locating of vehicles. Firstly, we designed a 79GHz (78-81GHz) Ultra Wide Band（UWB）millimeter wave radar, which can obtain the point cloud information of road edge features such as guardrail or green belt and so on. Secondly, the pixel semantic information of the drivable area of road can be obtained by the pixel semantic segmentation of image information through deep learning. Then, the road boundary equation in vehicle coordinate system is obtained by clustering and fusion of the road boundary point cloud information and the boundary semantic information of the drivable area. Finally, combined with vehicle (Inertial Measurement Unit) IMU information, the relative transverse and longitudinal distances between vehicles and road boundaries are obtained. The data acquisition test is carried out on real road scenes and compared with the detection results of lidar. Experimental results have substantiated the effectiveness as well as robustness of the proposed method. In conclusion, the method proposed in this paper has good robustness, economy and application value.
Jie Bai, Sen Li, Jinzhu Wang, Libo Huang, Lianfei Dong, Panpan Tong