Robot trials aim to cut strawberry harvesting time and control costs

Robot trials aim to cut strawberry harvesting time and control costs

Angela Youngman

Strawberry picking is labour intensive while the strawberries are fragile and easily damaged. Finding innovative solutions to this situation, especially with regard to the consequences of Brexit, is extremely important to the future of the sector. PBUK reports.

Robotic strawberry harvester

An entrepreneurial Cambridge-based, start up is aiming to provide that solution. Dogtooth Technologies has recently secured £60,000 funding from the Eastern Agri-Tech Growth Initiative to develop robotic strawberry harvesters. The grant will fund the development and implementation of the project.

According to Ed Herbert of Dogtooth Technologies, “The project is crucial in developing a cornerstone of our scalable strawberry harvesting system. Our system will allow growers to regulate costs, predict and increase yield through early intervention, and deliver a consistent, high-quality product. This product will accelerate the delivery of robotic harvesting and Dogtooth looks forward to helping our customers compete in an increasingly demanding market.”  

The innovative machine is designed to work within the narrow rows of polytunnels. It is set on caterpillar tracks so that it can move smoothly along the rows. The machine automatically looks at each strawberry as it moves along, utilising a memory bank comprising thousands of images in order to decide whether or not the strawberry is ready for picking.

If the strawberry is ripe, the robot arm extends, picks the fruit and places it in a punnet. All the punnets are contained on racks behind and in front of the arm and the robot assesses its memory banks in order to judge when a punnet is full.

At present the harvesting process takes 12 seconds per strawberry, but Dogtooth Technologies is planning to cut this time in half during the commercial trials taking place this year.

The first phase of the project, named Cabeirian, is now underway. Early stage trials have been encouraging with the robot being able to work quickly enough to provide a commercially viable alternative to human picking.  

Grower reaction

“Generally our offering has been met with massive enthusiasm from growers.  Recruiting enough picking labour is clearly an increasing challenge for British soft fruit growers, but the other benefits provided by our robotic picker are also of considerable interest such as the ability to deliver consistent picking performance for many hours, the ability to carry out grading in situ without repeated handling of the fruit, the ability to pick during the hours of darkness when temperatures are lower – maximising the shelf life of picked fruit,” says Dogtooth Technologies director, Duncan Robertson.

Ten harvesting robots are in the process of being manufactured. Each robot will be capable of navigating rows of strawberries, detecting and locating ripe fruit, before picking and checking the fruit condition and placing into punnets. The robots will be offered on a trial basis to UK growers and will work under the guidance of human supervisors.  

Future plans

If this initial trial proves successful, Dogtooth Technologies will expand production and seek to sell the harvesting robots to farms throughout the UK. Ultimately, the intention is to be able to develop the technology and the robots to such an extent that just one supervisor will be needed to deploy and manage teams of robots.  

Fully automated picking will become a reality, according to Robertson.

“We expect to provide commercially significant picking capabilities to reference customers in 2018 and to achieve widespread adoption by 2020.”

Who are the Dogtooth team?

Dogtooth Technologies is a start up venture involving a group of scientists. Co-founders are Ed Herbert, Duncan Robertson and Matt Cook.

Herbert is an engineer who has considerable experience in technology startups, research and development. Robertson has a PhD in computer vision from Cambridge University and has been involved in the commercialisation of cutting-edge machine learning and computer vision technology, as well as authoring numerous patents.

Matt Cook was part of the Microsoft Research team that developed body tracking software, and apart from being a highly skilled software engineer, he has considerable practical engineering skills and has a track record of managing the development of complex engineering projects. Dogtooth Technologies acquired its first funding in the form of a Technology Strategy Board SMART Award in September 2015.

Long term implications

The development of fully automated strawberry picking will have long term implications for the overall harvesting process as Robertson indicates.

“Among the interesting support requirements that will become necessary when deploying large numbers of robots onto farms and into polytunnels, transferring trays of picked fruit to distribution centres, gathering vast amounts of data obtained by robots to the cloud, and recharging robot batteries at the end of each picking session.”

“It’s impressive to see such ambitious and innovative development within the agricultural industry. To be able to offer this grant means we are supporting a movement towards the systemisation of the harvesting process, which will enable farmers and their teams to more accurately predict how long harvest will take,” says chairman of the Eastern Agri-Tech Initiative, Mark Reeve.

“The fruit-picking robots will allow farmers to buy a high quality and consistent harvesting capacity that will enable them to better control costs that can otherwise be fairly unpredictable.”

Robertson believes that robotic harvesting has the potential to make a major impact in more ways than one.

“We are a lot more than just a way of replacing human beings. Humans are remarkable, with so much dexterity. But what people like about our products is that they can do things humans cannot do,” he adds.

“It eliminates recruitment risk, gives a predictable picking capacity, with a consistent speed and quality. A robot does not get tired, and it still picks just as well after 18 hours of continuous use. Picking overnight is another huge advantage in terms of the shelf life of the product.”

“There is a Big Data opportunity too. Our machines will traverse each row of the polytunnel every two or three days and will take images and track the growth trajectory of each strawberry plant. The advantage of that in terms of yield and disease prevention is huge.”



The Latest from PBUK

Subscribe to PBUK!

Get regular produce industry insights, sign up for our email newsletter below.