Lightweight force-sensing tomato picking robotic arm with a “global-local” visual servo

Abstract

In this paper, a lightweight force-sensing tomato picking robotic arm with a “global–local” visual servo is designed to address issues such as large volume, large mass, high inertia, high cost, complex control and poor safety protection during automated tomato picking operations. Based on the relation between arm resistance and the joint motor current, a six degree-of-freedom (DOF) manipulator force-sensing system is established to develop a mechanism to ensure that human workers and plants do not accidentally collide with manipulators during picking operations. The system can determine the magnitude and position of the impact force during picking operations and immediately stops moving if the values exceed a certain threshold. A “global–local” visual servo is used to improve the repeated positioning accuracy of the end of the arm. A prototype is constructed for verification, and the results show that the average error of the force-sensing system of the lightweight manipulator in estimating the magnitude and position of the impact force is 7.6 %, and the average repeated positioning error of the global–local visual servo is 1.3 mm. When compared with a traditional interpolation trajectory planning algorithm, the error is reduced by 57.4 %, satisfying the accuracy requirements of automated tomato picking operations. During the picking process, the robotic arm can pick 93 % of the fruits with good safety protection, meeting the requirements for efficiently and automatically picking fruits and vegetables.