One of the biggest breweries worldwide contains approximately 550 employees at the Zoeterwoude brewery located in the Netherlands. HEINEKEN strains after a high-quality production flow where fully automation is desired. However, many employees have to perform physically labour. At the packaging area, feeding the unfolding machine is one of those actions in the process. The production line Colonne 43 has to produce 360boxes/hour. Which contains a box per Keg what has to be loaded.
This process takes place in the HEINEKEN factory located at Zoeterwoude-Rijndijk. The factory contains a brewing, packaging and storage area. Liesbeth Bommer, the manager of rayon 4, has detected an issue in her Colonne and translated this into a new project: “The pallet destacker”.
The Hague University of Applied Sciences tackled this project and transferred this to the SMR1 students. Project manager Danny van Egmond and the other students Julian Adelaar, Thijmen Adema, Saliem Amirkhan and Martijn Opmeer have taken on this assignment. During the 10 weeks, a solution is developed led by Thijs Brilleman.
Current situation
The unfolding machine needs boxes to work with. In the current situation, these boxes are manually placed by an employee. In packages of 14 boxes the pallet stacked. Between each layer (two packages) a sheet is placed which has to removed and placed into a trashcan. Each package has a weight of 15 kilograms. Because of the standard stacking pattern, the boxes are possibly 180 degrees horizontally rotated, which needs multiple rotations when placing boxes. These movements of the heavy packages can result in medical issues.
The current situation is shown in Figure 1. The delivered (red) pallets are stacked and needs to be emptied.
Our Goal
The goal of this project is to replace the employee and his tasks with reference to destacking the pallet. To reach this goal, a collaborative robot, vision system, PLC and HMI have to be used. The new system has to be able to operate on maximal work speed (300 boxes per hour) and fill the conveyor from an empty situation.
Solution
Multiple solutions were discussed to create a pick and place system for HEINEKEN. One of the limits of
the UR10 is the fact that the robot has a limited carrying weight. The UR10 robot that was used in
this project has a maximum carrying weight of 10 kilogram. The boxes were delivered in packages of
14 with a total weight of 15 kilogram. This exceeds the maximum carrying weight of 10 kilogram from
the UR10. Together with Heineken the decision was made that boxes would be delivered individually
on a stack.
End Of Arm Tool
For the end of arm tool (EOAT) a disc shaped frame was chosen (show in Figure 2). The circular shape was chosen due to the fact it offered more support than earlier versions. On the EOAT four 6mm connectors are mounted, each of which are connected to a check valve which is connected to a suction cup (shown in Figure 3). On the EOAT a proximity sensor is mounted which checks if there is a box underneath the EOAT. The 6mm connectors are connected in pairs to two aspirators and two pressure sensors. The aspirators are connected to the PLC of the robot. Suction cups are chosen and the robot is only moving one box at the time. Due to this solution, the robot only has 12 to 13 seconds to move a box onto the conveyor belt. This was initially not a problem until the speed of the robot became clear. This problem was eventually overcome by using acceleration and blending.
Vision system
For the vision, an Intel® RealSense™ SR300 was used. This camera is mounted on the top of the
frame as shown in Figure 2. By using OpenCV, a Python library, the system is able to detect boxes and
the sheets between the boxes. The camera is connected via a laptop to the PLC in the electrical box.
Push system
To make sure the boxes reach the unfolding machine a push system has been designed. The system is shown in Figure 4. It is also designed to prevent boxes from falling over on the conveyor belt. The push system is made of an electrical engine, gear, rack and a pusher. The pusher consists of a frame constructed of 25x25x2mm beams in the shape of the letter U. It also has a horizontal support beam. On that beam two brackets are mounted. These brackets provide additional support and a connection with the rack. The stationary electrical engine powers the gear. The gear interacts with the rack which will move forwards or backwards. The pusher is mounted on the rack using a slot and two support brackets. This system has not been built due to high cost.
Conclusion
The developed system is able to catch up the work speed. Though it is not able to overtake this. The speed of the robot is tested and pushed to the limit by introducing higher accelerations. These limits are set by testing the system and checking the safety during the movements. By looking more detailed after the limits, the robot might overtake the work speed.
To detect the boxes, a vision system with OpenCV is used. With this program is light incidence important to get the right results. The system makes use of HSV instead of RGB values. Despite of these values are less sensitive for light, the difference in daylight ensures issues. This problem might be solved when the system operates in the Heineken machine area. This area features constant light, which solves the issue.
The system is tested in an environment which is not equal to the HEINEKEN packaging area. The possibilities for using the right materials and sensors are too high in costs. To make sure the system will work in the packaging area, a 3D simulation is made. With this simulation all the necessary parts are visualized and developed.