Simulation As A Tool In Designing And Evaluating A Robotic Apple Harvesting System

Abstract

During the past several years, significant effort has been made to develop a robotic system to harvest fresh market apples in a modern apple orchard at Washington State University (WSU). This system has gone through several modifications to achieve enhanced system accuracy, speed, robustness, and reduction of fruit damage and cost. However, due to the short window of apple picking each year and the challenging environment of the orchard, those modifications had limited success in improving the robotic systems performance even with extensive laboratory testing before each field trial. To further assess potential problems, better prepare for the field trial, and make best use of the limited time in the field, in silico testing is necessary in addition to laboratory testing. In this work, computer modeling and simulation were used to help identify potential problem in the robotic system, evaluate the effectiveness of the newly made modifications, and aid in both algorithm development and configuration design of the system. By doing simulations with field study data and workspace analysis of the system, we developed a novel algorithm to improve the reachability of the picking arm. We also concluded that the overlap of the arm’s workspace and the fruit space is the key in deciding the reachability of the manipulator instead of the volume of arm’s workspace.