| Abstract (ENG): |
Industrial picking processes often depend on manual judgement to ensure the correct parts are picked, making them prone to error. While automated verification systems for picking exist, the cost to setup such a system often proves to be a huge bottleneck for startups and medium enterprises. This study introduces a real-time AI verification system that employes object detection to improve quality control in manual picking. The system combines a deep learning model You Only Look Once (YOLO), WebSocket-based streaming, and Redis messaging to provide fast verification of picked parts using only a standard webcam. An evaluation in a simulated industrial setting showed high classification accuracy, with processing times between 41–100ms per frame. |
| Citation: |
Siddiqui, Md Khalid and Hofmann, Benjamin and Engel, Tobias
(2025)
Development and Implementation of a Real Time AI-Based Part Verification System in Picking Process Using Low-Cost Equipment.
In: (Proceedings of the) 20th Conference of the Midwest Association for Information Systems (MWAIS). 17, May, 13-15, 2025, Stillwater, OK, USA.
|
