Dual Robot Coordination for Apple Harvesting

Abstract. A key task in robotic harvesting is collection of the picked fruit in a container. The usual collection method reported in the literature is a pick-and-place technique where the robotic harvester deposits the fruit in a stationary bin. This paper demonstrates a concept of dual robot coordination for fruit collection as an approach to potentially reduce the overall cycle time (average time per fruit) of robotic tree fruit harvesting. A kinematically redundant picking manipulator with eight degrees of freedom was developed for apple harvesting. Rather than displace the fruit to a collection bin, the robotic picking system, which includes the manipulator and picking end-effector, drops the apple into a robotic collection system near the point of fruit detachment. Cycle times of the pick-and-place and pick-and-catch methods were compared during a harvesting test in a simulated environment with an artificial apple tree located in a laboratory setting. The pick-and-catch harvesting method resulted in an over fifty percent reduction of average cycle time compared to the pick-and-place method. The robotic picking and fruit collection systems were then integrated with a machine vision system and briefly tested in a commercial apple orchard at the end of the 2016 apple harvesting season. Several observations from these preliminary field trials are discussed. Also, design modifications planned in preparation for more extensive field testing are reviewed at the end of the paper.