Vision-Based Autonomous Grasping Operations for Mobile Manipulator

Compared to CNC machine tools, mobile manipulators have the advantages of flexible movement, large working space, strong parallel coordination ability, and easy integration of multiple types of sensors, which can adapt to complex processing environments. This is gradually becoming a new trend of high-quality manufacturing of large complex components of major equipment. However, due to the large difference and irregular structure, selecting the grasping pose is difficult. Therefore, based on vision, this paper studies the grasping of a mobile manipulator for a variety of small aircraft structural parts, including target recognition, pose estimation and motion planning. By establishing our own data set, after identifying the specified grasping structure from multiple structural parts, the point cloud data of the structure part is directly processed to determine the optimal grasping position and orientation, and the mobile manipulator is controlled to realize the accurate identification and optimal grasping of the structural part in the real environment, which verifies the feasibility and effectiveness of the method.