Bolt anomaly detection based on yolov8 and point cloud

As a common connecting element, bolts are widely used in industries such as railway bridges, wind turbine towers, and building structures. However, due to long-term use, environmental changes, etc., loose bolts may cause instability and safety hazards to the equipment. Therefore, online bolt loosening monitoring is particularly important. For urban rail data, according to the box information, the template point cloud to be registered is obtained. The test point cloud to be registered will first be subject to yolov8 target detection, and then the IOU matching rule will be used to determine whether the bolt is lost, and the bolts that are not lost will be processed. Looseness detection, therefore, we directly take the target detection frame matched by the test point cloud and obtain the test point cloud to be matched. Point cloud registration uses GICP for registration. The target point cloud and test point cloud are segmented separately. Effective normal vectors are obtained for the target point cloud and test point cloud planes, and the normal vectors of the template bolt and the test bolt are obtained., calculate the angle between any two normal vectors. The angle between the normal vectors is used to determine whether the bolt is loose. The test data results show that this method can detect whether the bolt is loose.