Optimization Design of Intelligent Loading Initial Positioning for Cylindrical Workpiece s Based on Binocular Vision

In the intelligent handling process of Cylindrical Workpieces, binocular cameras are generally utilized for target recognition and pose detection. Addressing the issues of long operation time and a complicated workflow in the loading process, this paper proposes an optimization algorithm for the rough positioning stage. This algorithm employs binocular vision image processing technology for pose detection. The method can effectively eliminate noise caused by complex environments and correct image distortion using the binocular cameras' radial distortion model. It optimizes the pose detection pipeline of the intelligent handling system and incorporates a rotation function around the Z-axis during the rough positioning phase, thereby reducing the number of actions required in the subsequent fine positioning process. Experiments conducted on an existing platform verify its optimization performance. The results demonstrate that the optimized algorithm reduces the total time of the pose detection process by over 10 seconds. It also resolves the program crash issue caused by an excessive angle between the Cylindrical Workpiece and the parking rack during pose detection, thereby enhancing the intelligence and flexibility of the intelligent handling process.