Event Capture Systems (ECS) is a US company that offers manufacturers camera-based inspection systems for quality assurance - with the aim of speeding up processes while reducing rejects and downtime. ECS relies on artificial intelligence with the support of image processing provider Baumer and its 'AX Smart Cameras'. ECS is implementing the company's own AI software solution in Baumer's cameras in order to monitor processes in paper production, one of the company's focus industries. In the paper industry, where wood chips are processed into large paper rolls, quality inspection is a demanding task.

When introducing AI-supported inspection, ECS was faced with two challenges: First, the processes that would benefit most from AI had to be determined in order to develop customized software and supporting AI models. Secondly, due to the harsh environmental conditions and costly downtime in paper manufacturing, ECS needed solid hardware to run the software.

Inspecting raw materials and process steps

Thanks to the complete package of NVIDIA JetPack and GenICam-compliant Baumer camera software, customers can set up their applications easily and efficiently directly on AX Smart Camera.

© Baumer

The potential of AI is particularly evident in two areas of application in paper production:

• Classification of wood chips: the production of paper begins with wood chips as the raw material for pulp production. The purity of this raw material is essential for high paper quality; until now, it has been difficult to inspect automatically. The wood chips are processed into pulp, which is later turned into paper. Knowing the shape and quality of the fibers is crucial for efficiency in the process and for compliance with supplier quality standards. Based on the time of inspection, the raw material assessment can be transferred to the supplier (external or internal). At the mill, the actual quality parameters of the wood chips are compared with specified quality standards and corrective measures are defined to ensure optimum results when processing the raw material into paper.
• Detection of edge cracks: Another important task for the smart cameras is to inspect the processed paper webs for edge tears. If damaged edges are not detected quickly, this leads to rejects, reduced production output and increased downtime.

But what exactly is AI needed for in this application? It is important to look at how quality inspection in paper production works without AI-supported smart cameras. The first solutions at ECS for the inspection tasks of chipping and edge cracking were based on conventional, passive image processing cameras. These send data to a computer - with the camera acting as a sensor that sends the data to the application on a server computer. This solution worked well, but the installation costs for data transmission from the sensor to the computer and the associated requirements proved to be an obstacle to its use. For example, laying signal cables and additional backbone cables often requires considerable planning, time and the deployment of project engineers directly on site. In addition, the solution with conventional passive cameras involves three components that need to be maintained or require troubleshooting: The maintenance engineer has to look after both the camera and the signal backbone as well as the computer for image processing. Compared to this, the edge computing function of the Smart Cameras ECS offers the following advantages:

• The smart cameras are a solution with 'out of the box' functionality, minimizing installation costs and complexity.
• As a centralized solution, they provide actionable data for quality control - as opposed to a distributed architecture that results in higher maintenance costs in the long run (assuming someone is available for maintenance).
• Easy maintenance: If the screen goes 'black', a new AX camera with implemented application software is simply installed.

Expandable status quo

The author: Holger Thissen is PR Manager at Baumer in Frauenfeld (Switzerland).

© Baumer

According to the United States Census Bureau survey of 2023, 3.8 % of US companies currently use artificial intelligence to produce goods and services; the Bitkom survey 2023 indicates that 15 % of German companies use AI.