14-04-2023 дата публикации
Номер: CN115965960A
Принадлежит:
The invention discloses a weld joint identification method based on deep learning and 3D point cloud, and the method comprises the following steps: S1, presetting a camera shooting position, controlling a robot to a corresponding position through a mode that eyes are on hands, shooting point cloud through a camera, and converting the point cloud data from a camera coordinate system to a mechanical arm base coordinate system through hand-eye calibration; s2, point cloud reconstruction is carried out through a voxel probability model based on deep learning, weldment complete point cloud data is extracted to carry out preprocessing on point cloud, and sampled point cloud is obtained; s3, identifying a welding seam, segmenting a plane through a pointnet + + algorithm, and searching based on distance features to obtain welding seam point cloud data; and S4, linear fitting is conducted on the weld joint feature points, welding pose vectors are calculated, weld joint information is sent to the ...
Подробнее