BEV Representation
Bird's-Eye-View (BEV) representation is a crucial technique in autonomous driving, aiming to transform multi-sensor data (primarily camera images) into a top-down 2D view of the surrounding environment. Current research focuses on improving BEV generation accuracy and efficiency through various approaches, including transformer-based architectures, polar coordinate systems, and innovative fusion strategies for camera and LiDAR data, often incorporating learned priors or masks to enhance feature extraction. These advancements are significantly impacting autonomous driving systems by enabling more robust 3D object detection, HD map construction, and place recognition, ultimately enhancing safety and reliability.
Papers
CRPlace: Camera-Radar Fusion with BEV Representation for Place Recognition
Shaowei Fu, Yifan Duan, Yao Li, Chengzhen Meng, Yingjie Wang, Jianmin Ji, Yanyong Zhang
Infrastructure-Assisted Collaborative Perception in Automated Valet Parking: A Safety Perspective
Yukuan Jia, Jiawen Zhang, Shimeng Lu, Baokang Fan, Ruiqing Mao, Sheng Zhou, Zhisheng Niu