Design of PLC-Integrated Object Detection for Categorizing Conveyor

The purposes of this research aim to retain an artificial intelligence model using the NVIDIA Jetson Nano Developer Kit as AI module, integrated the conveyor belt sorting system and investigate accuracy and the optimization of velocity belt. The essential components of the sorting system comprise three parts: the detecting module, which includes an image sensor was CMOS 2 MP pixel and an object detection image processing device; the automatic control system, which comprises a programmable logic controller (PLC), a pneumatic system, control valves, conveyor belts, motor control system, and the main computer system for internal communication and presentation of system data through the Monitoring Dashboard. Object detection models were re-trained model after the dataset was prepared by capturing and labeling images. The AI module and automation conveyor are linked through MQTT, MC protocol, and OPC server. Kepwear was installed on a host computer to communicate and transfer data messages. The results of research show that the conveyor's velocity belt speed was 0.11 m/s, with an average correctly detected value of 82.50 percent and a low-high detecting accuracy between 75.00 and 91.7 percent. At a speed of 0.22 m/s of the conveyor's velocity belt speed, the greatest value is 83.33 percent, its lowest value is 66.67 percent, and its average detected value is 75.00 percent.