To enhance the accuracy and efficiency of soft fruit sorting processes, this study presents the design and implementation of a Delta parallel-arm soft fruit sorting machine based on improved Object detection algorithm and Genetic Algorithm (GA). By incorporating the Convolutional Block Attention Module into the Object detection algorithm architecture, the feature extraction module was optimized, thereby improving detection accuracy and robustness for fruits with varying maturity levels and damage degrees under complex background conditions. At the actuator level, GA was employed to globally optimize the grasping and placement trajectories of the Delta parallel arm, combined with gradient acceleration and deceleration control to apply smooth constraints on motion velocity and acceleration. This approach maintains operational cycle time while minimizing mechanical impact and vibration, consequently reducing potential damage to strawberry epidermis and flesh. Experimental results demonstrate that the proposed sorting machine achieves superior performance in target detection accuracy, sorting efficiency, and fruit integrity rate during the sorting process compared to conventional methods, exhibiting overall stability and feasibility. The outcomes of this research provide an efficient and precise sorting solution for smart agriculture and offer substantial support for future research and applications in automated agricultural product sorting technologies.
GUO et al. (Sun,) studied this question.