Research on vehicle target detection based on FMCW millimeter wave radar and vision fusion

With the rapid development of intelligent driving technology, how to ensure the stability and high reliability of the system in a complex and changing environment has become an important issue that needs to be solved urgently. The performance fluctuations of a single sensor in different environments, especially in extreme weather conditions, often limit the accuracy of target detection. To address this problem, this paper proposes a target detection method based on the fusion of millimeter-wave radar and visual sensor to give full play to the complementary advantages of the two. The visual part uses the YOLOv5s algorithm for efficient target detection; the radar part removes invalid point clouds by filtering to accurately identify valid targets. On this basis, this paper designs an adaptive fusion strategy: decision-level fusion is used under normal weather conditions to make full use of the high resolution of the visual sensor; data-level fusion is used under extreme weather conditions to give full play to the anti-interference ability of the millimeter-wave radar. Experimental results show that the fusion model performs well on the NuScenes dataset. The detection accuracy rate reaches 87.6% under good weather conditions and 76.9% under bad weather conditions, which verifies the effectiveness and adaptability of the method.