A lab-customized autonomous humanoid apple harvesting robot

This paper presents an autonomous humanoid robot, namely, LABOR, for apple harvesting. Comprised of a binocular camera, a humanoid dual-arm operating system, and a mobile vehicle platform, the devised robot is capable of identifying, positioning, grasping, and picking up apples via its vision identification and dual-arm harvesting systems. Both the robot structure design and the functional systems, as well as their implementations are described in detail and were verified via experiments in a simulated orchard environment. Results show success rates of 82.5% and 72% for the apple recognition and harvesting functions respectively, and the average apple harvesting time was approximately 14.6 s. The wide flexibility and high scalability in the robotic structure, dual-arm coordinate control, and independent harvest planning may facilitate its application in agriculture for the harvesting of other fruits, such as tomatoes, pomegranates, and pears.