Student develops PotatoBot to minimise damage planting Jersey Royal potatoes
Due to their fragile paper-like skins, Jersey Royal potatoes must be planted by hand to minimise damage, but with the steep fields in the country this can be an arduous task. With labour difficult to find, switching to a mechanised method to plant the seed potatoes has not been successful, until now, with the introduction of the PotatoBot.
For the past three years PhD student Elijah Almanzor has been developing the PotatoBot robot as part of his studies in the Bio-inspired Robotics Lab (BIRL) at the University of Cambridge, supervised by Professor Fumiya Iida. This machine has been designed to help plant Jersey Royal potatoes on the island of Jersey, which exports up to 30,000 tonnes of the crop per year.
Although there is still some work to do before the unit can be fully commercialised, Elijah’s creation will be a huge help to Jersey potato farmers, and its technology could be transferred to aid other crops production. PotatoBot is essentially an electric robot which can be powered by a generator in the field or from a tractor, with the automation powered by AI computer vision. The computer vision uses a neural network for detecting and segmenting the Jersey Royal’s given a depth camera view on the robot’s wrist.
The purpose of the PotatoBot is to plant Jersey Royal Potatoes in a similar manner to human planters
Based on this, the PotatoBot can then localise the potatoes from the Jersey boxes for picking and planting operations, regardless of their placement and position within the box. Elijah said: “My PhD is supported by the Agriforwards CDT, a doctoral training programme dedicated to developing AI-driven robotics for agriculture. “The purpose of the PotatoBot is to plant Jersey Royal Potatoes in a similar manner to human planters. Unlike other varieties of potatoes, Jersey Royals necessitates manual planting due to seed potato shoots which can easily be destroyed by mechanical planters.”
Text continues below picture
Robotic method of planting
The quest to invent a robotic method of planting the potatoes came after alternative mechanical methods damaged the crop in the past. The PotatoBot needs a performance of 3.5 second cycle time with a 90% success rate for it to be commercially viable. It also needs an autonomous mobile platform such that it can be deployed at a field.
The project is generously supported by the UKRI EPSRC and the Jersey Farmers Union. The robot is currently at a proof-of-concept stage. Elijah added: “It needs more development with field-testing before it can be commercially available. The newer laboratory iteration with a much faster type of robot arm (SCARA robot), however, is close to our goals.
Text continues below picture
Minimise costs
“Currently, we are unsure of how much the price will be commercially, however, I’ve made sure to use inexpensive and off-the-shelf components to minimise costs, as agriculture in general has tighter revenue margins compared to other industries. “Due to this project, I’ve formed a Robotics as a Service start-up (Robotlux) with fellow lab members, with the goal of solving real world agricultural problems, as there is a genuine need to increase food production efficiency given an ever-growing human population.”
“Hence, hopefully with a bit more development, we can transfer the PotatoBot from the lab to the real world,” he said. Elijah also pointed out that the AI driven system for manipulating agricultural produce used in PotatoBot, is easily transferable to other vegetables with some hardware changes.
