EPFL has developed a robotic arm capable of identifying ripe fruit and can harvest them without damage. (Image credit: EPFL)
Engineers from EPFL's CREATE Lab have developed and tested a robotic arm capable of harvesting raspberries without damaging the fruit in the process. Raspberries are delicious but delicate, as the fruit features dozens of druplets with thin skin that can get damaged just by handling them. As the world grows, the demand for food has increased, and farmers are looking at ways to maximize food production without increased costs. To that end, farmers are looking to robots to perform manual labor, including picking fruits and vegetables.
Harvesting food isn't as easy as it may appear. Some fruits and vegetables have a thin outer layer, such as tomatoes and raspberries, so harvesting requires some level of finesse to prevent damage. Humans have a leg-up in this area as our sense of touch can tell us if the fruit is ripe and what amount of force is needed to remove the fruit from the stem. The engineering team from EPFL designed a robotic arm that can identify and harvest ripe fruit by allowing the robot to learn to pick fruit in a laboratory setting.
The engineers designed the robot using cameras and machine learning to identify the ripe fruit and graspers to harvest them. This was done using silicone representations of raspberries that look and feel almost like their real counterpart.
After successful testing in the lab, the team then tested the robot in the field with natural fruit; however tests showed the robot was subpar compared to humans. The robot managed to identify and harvest 60% of ripe raspberries, while humans achieved 90% accuracy. While the robot was outperformed, the engineer's state improvements could raise that percentage to match humans with later revisions.
Have a story tip? Message me at: http://twitter.com/Cabe_Atwell
