Research

(1) Shop-floor monitoring systems based on vision data 

• Objectives: (1) Implementation of a miniature-size real shop-floor model; (2) development of vision-based real-time multi-object tracking for shop-floor logistics; and (3) development of an automatic generation method of discrete event models based on vision data

• Motivation: As a step towards Industry 4.0, sensor-based monitoring systems have been broadly applied to manufacturing sites. However, it is challenging for small and medium-sized enterprises (SMEs) having the limited investment to adopt the entire system including the hardware and installation. Vision-based monitoring systems can be applied to overcome the practical limitation.

• Research opportunities: (1) How to improve the accuracy of vision-based multi-object tracking; (2) how to optimize the number, location, and angle of cameras; and (3) how to model finite state automata based on vision-based unstructured data

• Requirements: Great Ball Contraption (GBC) modules, object detection models (YOLO, OpenCV, etc.), Raspberry Pi with a camera module, and Tensorflow

• Contributions: The vision-based monitoring and management systems can support SMEs by facilitating the collection of shop-floor data. This could serve as a foundation for SMEs to take a step closer to their smart factories. 

(2) Part Decomposition for Additive Manufacturing