Harvesting soft fruit is among the hardest tasks to automate: a gripper must sense ripeness, grasp without crushing and detach without tearing. Researchers in Rob Shepherd's Organic Robotics Lab at Cornell Engineering, with horticulturist Marvin Pritts of the College of Agriculture and Life Sciences, have built a soft robotic hand that does all three, described in Nature Communications. Strawberries served as the model crop.
The design threads stretchable fibre-optic sensors through the gripper itself: one measures each finger's curvature, another the pressure at the fingertip, letting the robot estimate a fruit's shape and stiffness and adjust its grip accordingly. Because the sensors share the gripper's mechanical properties, the flesh of the hand effectively feels the fruit. A camera in the palm locates berries hidden by leaves, and a planetary gear mechanism rotates the wrist so the fruit is twisted off its stem rather than yanked.
The team validated touch-based ripeness predictions against colour and used stiffness measurements to estimate harvest timing. Beyond strawberries, the researchers point to fruit whose ripeness is hard to judge by eye or that ships poorly, such as avocados, pineapples and pawpaws. They also envisage fleets of smaller robots enabling mixed cropping of mutually supportive species, reducing pesticide and fertiliser needs. The work was funded through the NSF Center for Research on Programmable Plant Systems and the Cornell Institute for Digital Agriculture.