Innovation in non instrusive data extraction
Shop floor data acquisition challeges
One of the biggest challenges in capturing data from the production floor is acquiring the necessary information from machines that are actively running. The most effective approach is to use high-level communication via standard protocols such as Modbus or other PLC communication protocols. However, in many cases, the required data is either unavailable or originally retained by the machine’s vendor.
Moreover, some machines cannot afford to be shut down due to high production demand.
Look to the right for a solution ➝
QuThe solution :-
Non Intrusive Edge AI Data Extraction
Today’s industrial challenges require agile solutions. Our cutting-edge data extraction technology lets you monitor progress seamlessly, ensuring enhanced productivity without disruption.
“WTF EYE” – What The Fish Eye is a computer vision AI solution developed by the AIoTmission AI Team to address a common challenge on the production floor: machines that cannot be stopped and where traditional data interfacing methods are not feasible. Despite this, Industry 4.0 teams still need real-time machine data for OEE (Overall Equipment Effectiveness) tracking.
To meet this requirement, a compact edge device and an IP fisheye camera are used. The camera is mounted above the machine to avoid obstruction from moving parts, providing a clear view of the stack light (also known as a tower light), which indicates the machine’s operational status.
The Axiomtek Jetson Edge AI platform plays a crucial role, offering low-power yet high-performance capabilities to perform on-site inferencing efficiently.
Special thanks to the AI team for training a model capable of recognizing the various stack light indicators that reflect the machine’s status—such as green for “running,” red for “stopped,” and yellow for “in process.” With AI inferencing done directly on the edge, the machine status can be visualized on the AIoT Edge Connect Cloud Dashboard throughout the shift.
One of the key features of this system is its ability to monitor up to four stack lights simultaneously by intelligently partitioning the fisheye IP camera’s field of view.