Automated quality inspection in the food industry | Stäubli x Bolletje

Challenge

AI-powered, robot-assisted quality inspection of baked goods

Bolletje, a Dutch baked goods manufacturer, has automated its quality inspection. An inspection cell consisting of a camera system and a four-axis robot enables the inspection of up to 1,200 rusk slices per minute. The image data is analyzed using AI and used to optimize the system. A Stäubli TS2-80 SCARA robot sorts out defective products.

How do you check the quality of 1,200 slices of rusk as they come off a 200-meter-long oven line every minute? A team of five employees is assigned to sort out non-conforming products—such as slices that are too browned—using their trained eyes and quick reflexes. This is how Bolletje in Almelo, the Netherlands, operated for many years.

Solution

Investment in AI-powered quality control

Today, things are different: A compact robotic cell consisting of a camera, a four-axis Stäubli robot, and an AI-based IT platform handles the task. The rusk slices are captured by a camera, the image data is analyzed within milliseconds, and the products classified as “non-conforming” are sorted by the Stäubli robot.

The data from the 100% inspection is thoroughly analyzed. Lo Huls, COO of Bolletje: “We record the type of deviation and link it to the system data. This task is performed by our data analytics tool, which monitors all ovens and other process steps. This enables us to identify the causes of quality defects and implement targeted corrective actions.”

Bram de Vrught, Managing Director of systems integrator QING Food Automation, describes the practical implementation of this process: “The system captures the images, transfers them to the STAQ platform, and classifies the products and various defect patterns. We can then train the AI based on this classification. Overall, the system is very user-friendly, allowing companies to deploy it independently and extend the technology to other products or new quality criteria.”

From the very beginning of STAQ’s development, QING relied on four-axis robots from Stäubli. A delta robot would have required more installation space, increasing the cell’s overall footprint. Under these conditions, the highly dynamic Stäubli TS2-80 delivers the best performance.

Results & Customer Feedback

• The inspection system, which uses a ceiling-mounted robot, checks 1,200 slices of rusk per minute.
• The compact system is integrated into one of Bolletje’s crispbread production lines and can be moved to another line if necessary.
• A specially designed needle gripper picks up slices of rusk for sorting.
• The human-machine interface of QING’s STAQ platform makes it easy to use.
• The robot can sort up to 80 slices of rusk per minute.

Compact solution with high efficiency

Automation has allowed the five Bolletje employees who were previously responsible for visually inspecting 1,000 to 1,200 rusks per minute on this line to take on other tasks in the industrial bakery.

The use of Stäubli’s VALtrack software also proved to be particularly advantageous. It synchronizes the robot’s movements with those of the conveyor belt, fulfilling a key requirement for fast and precise gripping of the products being sorted.

“We could have implemented a system with two robots. However, this would have nearly doubled the costs and space requirements, and coordinating the robots would have significantly increased the programming effort. Therefore, a robot optimized for peak performance is the more economically sensible approach. The TS2-80 continues to operate within its design limits, so we expect a long service life and minimal maintenance requirements even during 24/7 operation.”

says Bram de Vrught, Managing Director of systems integrator QING Food Automation

Source: Stäubli https://www.staubli.com/global/en/robotics/industries/food/bread-and-baked-goods/quality-inspection-of-baked-goods.html