Fieldwork wins UK government grant for the BerryAI Project

Photo: Fieldwork
Photo: Fieldwork

Fieldwork Robotics Ltd. (‘Fieldwork’), developers of selective, adaptive and modular harvesting robots, announced that it has been awarded a £600k government grant by the Department for Environment, Food and Rural Affairs (DEFRA) and Innovate UK.

The BerryAI Project will support the development of Fieldwork’s technology, bringing AI-powered vision and advanced decision-making technology to Fieldwork’s robots. The BerryAI Project will underpin the development of Fieldwork’s next autonomous model and has two aims: to enhance the AI-powered vision of the robot, and to improve its decision-making capabilities.

The updated robotic system will have enhanced artificial intelligence (AI) capabilities, meaning that the model will have the capacity to work in a fleet with one operator running multiple robots across the field. This significantly reduces the labour intensity of the harvesting process and improves harvesting efficiency.

Super-human vision

Until now, the cameras on Fieldwork’s robots have mostly utilised the human visual spectrum, mimicking the ability of a human harvester. This project however will incorporate technology that utilises wavelengths invisible to the human eye, facilitated by our partner Fotenix, leading developers of crop analytic software.

This development will improve the robot’s ability to detect crop and determine ripeness. Development of these two functions will make the robot more autonomous and more efficient as an alternative to human harvesting, a key step towards Fieldwork Robotics’ goals.

AI-powered decision making

In addition to improving the super-human vision of the robot, the BerryAI Project is also targeting decision making within the robots in order to improve their autonomy. The expectation is that improved AI decision making will allow the robots to operate for long periods of time with minimal human oversight, making the robots increasingly cost-effective and efficient. It will be increasingly realistic to manage a large fleet of robots with a small workforce.

Ed Asscheman Online editor Future Farming