Zwethlande is a grower of Ficus plants. They have over 30 years of experience with this type of plant, which explains their expertise. At some point they decided to use this experience to grow other types of plants as well, like the Yucca. This plant is grown from a stem, with a variable amount of heads. The project is designed for the Yucca plants.
In the current situation the pots are taken from the tidal section (a section with high and low water for watering plants) with a forklift. The forklift places the plants on a big conveyor belt, working as a buffer. The plants are pulled on another conveyor belt by a separating slider. On this next belt, the pots are sorted by height and sent to a new separating conveyor belt. The height of the plant will be checked by infrared (IR) sensors. These sensors are placed above each other, with a set distance in between.
After the sorting, the plants will be pulled onto another conveyor by a grabber. This grabber will get a signal from the sensors which buffer it needs to go to and it acts accordingly. The plants are not sorted on the amounts of heads, but this program is already made by a previous SMR-group (https://smrdelft.nl/plant-sorting-system/). After the sorting, Zwethlande employees will put the plants into trays, which are placed on a trolley afterwards. While filling, the workers sort the plants on the different types of heads.
At the moment this system doesn’t work properly. Some of the current problems are: The plants fall of the conveyor belt when they are moved, the work needs a lot of human interaction, it is a boring job for the employees and every time an employee needs to check if the system worked properly. For these problems, Zwethlande wants a solution.
How did you solve this problem?
A concept to improve the whole system has been made. This system consists of multiple conveyor belts working as buffers and transportation belts. This system is too big to design and build in the time given for this project. Because of this, only a part of the system is fleshed out. The former project has been taken into account as well. The main focus of the project was the pick-and-placing of the pots into the trays. The system is designed for two different kind of trays, 2×4 plants and 2×5 plants. The pots are counted by an IR sensor, so the robot knows when to pick it up. The system detects when a pot is in the EOA tool, which then slides to the side. This way the position of the plants is fixed, which is needed for the placing into the trays. Depending on the type of the pot, the EOA tool is filled with either four or five pots. When the tool is filled, the robot will move the plants to the tray.
Before the plants can be placed into the trays, the system must know if there is a tray at the placing point. If there is a tray, the filled EOA tool is lowered and the pots are placed in the tray. This sequence is repeated with the same amount of plants twice. This is necessary because a tray for big pots contains 8 (2×4) plants, the small tray contains 10 (2×5) plants. After the tray is filled, it will be checked by cameras with vision software. These cameras check if the plants are placed correctly in the trays. It works by determining the angle of the stem of the plant. The plant is given a certain threshold from the horizontal axis (x-axis). If the angle is not within the threshold, the plant is not placed correctly and there will be a False signal. This signal can be used in different applications.
What were the big decision you made?
The first concept in this project was a sliding mechanism to position the plants for the EOA tool. This should shorten the total time needed to fill a tray. Because those mechanisms are quite expensive, the group made its own. This took too much time and so we decided to disregard it and positioning the plants with the EOA tool itself. If we designed the slider differently from the beginning (we used a trapezium rod instead of a belt drive, which slowed down the speed dramastically).
By accident we found out how we could alter the images shown in the vision software. When the group was testing the vision software, it didn’t recognize the plants. When somebody accidently passed walking, wearing a red sweater, the system started detecting the plants. Because of this, the group found out that there had to be a red background.