抄録
A stationary robotic strawberry harvester, combinable with a movable bench system, was developed. Initially, the features of strawberries, such as the visibility of fruit, the picking force required, and spectral reflectance were investigated. Subsequently, a machine-vision system was constructed, comprising units to measure position and coloration respectively. Considering the visibility of strawberry fruit, these units were optimally located to detect mature fruit and estimate their coloration. White, red, and green light-emitting diodes were installed for more accurate coloration measurement. The average error was 5.4%, and its standard deviation was 10.8%. We developed an end-effector with three functions: 1) to prevent the fruit adjoining the target fruit from being touched during the approach, 2) to force adjoining fruit away from the target fruit during the picking motion, and 3) to remove peduncles from harvested fruit, as done manually. The performance test confirmed that the stationary robotic harvester outperformed the conventional robots, namely, the harvester achieved a stable high harvesting success rate of 67.1%, and was able to successfully remove the peduncles from 88.0% of the harvested fruit.
