Research on Intelligent Sorting System
Yanghong Mao, Chuixin Chen · 2022
In order to achieve the goal that the manipulator can automatically obtain any position within its working range, and can adjust the end actuator for grasping, a vision sorting system is designed. The system passes through image processing, target recognition and positioning, and finally realizes the sorting effect through end control. The acquired image is enhanced and edge detected. Aiming at the problem that the deep learning algorithm takes a long time, The traditional image processing algorithm and deep learning target detection algorithm are organically combined to reduce the use delay of image processing in the system. Because the training of deep learning model requires a large number of data sets, but the amount of data in manufacturing factories is small, and data labeling is time-consuming and laborious, YOLO v3 algorithm is used to improve the efficiency of training model under a small number of labeled data sets. In the algorithm of position, the centroid is extracted as the positioning reference coordinate, the kinematics model of the manipulator is established. and sent to STM32 through the serial port. Then the STM32 completes the control command to grasp and place the workpiece and complete the sorting task. The experimental results show that the system meets the design requirements.