Students: Peter Verweij, Pascal van der Krogt, Hugo Vader & Alex Lorier. (Spring 2016)
Bejo Zaden B.V. is a company that is specialised in the production, processing and upgrading of vegetable seeds. Bejo has more than 100 years of experience in the seed industry and is amongst the leading specialists in breeding, production, processing and sale of quality seeds. The company is based in the Netherlands at Warmenhuizen in the so-called Seed Valley.
The project consists of a packing machine that processes cans with delicate and expensive seeds. The empty cans go into the machine, are filled up with seeds and then closed by the machine.
- At the current production line the cans are manually moved from a pallet into the system.
- Subsequently the cans are filled.
- Then the cans are closed.
- Thereafter the cans are weighed. The cans with the wrong weight will be removed.
- Afterwards the cans are labeled
- Finally the cans that come out of the conveyer upside down are packed. (This happens manually right now, but that’s what will be automated)
Our task was to automate the last part (number 6) where the cans go into boxes. Because of the repeating procedure and constant stress on the employee’s body, that might even violate the ARBO laws one day, the company wants to remove the manual labor from the production process. The company will not fire these employees because they will still be needed to start and monitor the machines. They only need a basic training on how to work with the robot. Currently only one employee is working at this particular production line. Due to an expected growth in line usage the company is planning ahead to already automate the line.
Problem Definition and Current situation
The goal of our project was to design a robot which can pack cans and trays into a box and check if this was done correctly with vision. Our scope starts at the input of the labeled cans and ends with the filled, but not closed, box. To help structure this process we divided the problem into subcategories:
- Picking up the cans;
- Placing the plastic can holder;
- Check if the cardboard box is properly filled with cans;
- If one can is missing the robot has to be able to recognize and place the missing can.
For this project we have used the Kawasaki FS06L industrial robot and a Cognex 7200C camera for the vision. We decided to use a pneumatic system for both the trays and the cans. We designed a multitoolhead where the cans and trays are picked up on the same side. The suction cups are integrated in the flat surface where the cans are picked up. When the cans are picked up the suction cups don’t touch the cans, this is necessary to get a good vacuum to pick up the cans. The suction cups have to stick out a little bit to pick the trays.
Figure 1: Pneumatic end of arm tool concept
Figure 2: Pneumatic end of arm tool concept
Because the cans are coming in upside down we designed a slide where the cans are turned. This slide makes sure that the cans are always orientated in the same way and positioned at the same location. This makes the system more reliable. At Bejo they use some different sizes of cans, and all those cans have to be turned. This makes it a lot harder to make a slide where both, the big and small cans can fit in. This is why we designed a part that could be clicked on the slide so it is suitable for the small cans as well.
Initially we designed a magnetic gripper to grab the cans and a pneumatic end of arm tool to grab the trays. However, when we started testing the prototype we found out that the cans weren’t magnetic enough to be lifted by our electromagnets. This was a big setback because we tested this principle with permanent magnets and that worked great. This setback is why we tried some new ways to pick up the cans. This is how we came tot the pneumatic system.
Figure 3: End of arm tool with electromagnets
Implementation & Execution
All the parts we designed were made during the first 6 weeks. Everything was build at the Beta Factory and integrated to our simulation at school. At school we made a test set-up to simulate the production line of Bejo. In the last two weeks we were busy with writing our program and busy to make everything work to go testing at the factory itself.
Figure 4: Bejo truck at The Hague University
The integration on the factory floor went very well. When we came to Bejo we could immediately start with installing our robot. We had all the space we needed and could ask for everything if we needed something. When placing the robot and our slide we decided to extend our slide so we had a better point for picking up the cans. The next day we had a whole day to program and test our solution.
Figure 5: The Kawasaki being installed at Bejo
After the entire day of testing we came to the conclusion that the reliability of our slide wasn’t satisfying. There were to many movements in the slide so the cans weren’t always on exactly the same place to pick them up. While the other parts worked really well.
To conclude we can say that the solution we came up with wasn’t reliable enough. If you look to the parts apart of each other, there are enough parts that are reliable. These parts could be used to get it working together with some new components.
The turning of the cans was in our case the hardest part. In the near the future the machine where cans come out will be changed, and the cans will not be fillet upside down anymore. This means that the unreliable part of our solution will be eliminated of the system, and it should be possible to make it work with solution we came up with.