Comparative study of mechanical damage caused by a two-finger tomato gripper with different robotic grasping patterns for harvesting robots

The fragile structure of the tomato fruit body leads to susceptibility to bruising caused by the aggressiveness of harvest and postharvest processes. Thus, grasping without damaging the tomato fruits is a key barrier to the replacement of manual labour by robotic harvesting. In this study, a four-element Burger model was used to express reversible viscoelastic behaviour and deformation characteristics of tomatoes at early and middle red-ripening stages. Additionally, creep tests were conducted to obtain the viscoelastic parameters of the Burger model. The model for plastic deformation of tomato during grasping was finally developed based on input force, contact time, and viscoelastic parameters. In order to explore the least damaging grasping pattern, plastic deformation caused by three grasping patterns (denoted as Pattern I, Pattern II, and Pattern III) were investigated and compared in our study. A linear function, a Butterworth amplitude square function, and an exponential function were used to represent the velocity variations in the three grasping patterns during the robot grasping operation. This was used to solve the model of plastic deformation of tomato, and the changing rules of tomato plastic deformations under different grasping patterns were analysed under constant grasping time. The results indicate that grasping Pattern III is the optimal grasping strategy, the lowest plastic deformation of tomatoes is obtained with grasping time t0=1s and grasping velocity v0=1mms−1 and the plastic deformations correspond to 0.0026 mm and 0.0098 mm for tomatoes at early and middle red-ripening stages, respectively. A grasping control experiment was also conducted under grasping Pattern III, and the correlation coefficient of 0.99 for the simulation and measured results indicated the rationality and feasibility of grasping Pattern III as the optimal grasping strategy. Our study provides a theoretical basis to optimise agricultural robot grasping.