A methodology of orchard architecture design for an optimal harvesting robot

To improve robot performance for agricultural tasks, and decrease its cost, the robot can be optimally designed for a specific task in a specific working environment. However, since the environment defines the robot optimal kinematics, the environment itself should also be optimised for optimal robot performance. The objective of this paper is to present and demonstrate a methodology for simultaneous optimal design of robot kinematic and the working environment. This methodology was demonstrated by an example on a tree orchard design for an apple harvesting robot. First, an optimal robot structure for apple picking task was found for a number of tree architectures (shaped by different training systems): Central Leader, Y-trellis and Tall Spindle. Results indicate that for minimising the average apple picking time, the Tall Spindle architecture is preferable for the robotic harvesting of both a single tree and a tree row. Further, the influence of the robot platform motion time on the chosen robot kinematics and the tree training system was analysed. Results show that for fast platforms, the Tall spindle architecture is advantageous. If the platform movement between positions near the trees is slow, the Central Leader architecture is favourable. Additionally, the tilt angle of the Y-trellis training system was analysed using simulated models created by the L-systems simulations. The optimal tilt angle was found to be nearly horizontal (85°), allowing the robot designer to choose the optimal combination of the robot kinematics, number of robot harvesting positions around the tree and the tree training system.