Stable, efficient and lossless fruit picking has always been a difficult problem, perplexing the development of fruit automatic picking technology. In order to effectively solve this technical problem, this paper establishes a multi-objective trajectory model of the manipulator and proposes an improved multi-objective particle swarm optimization algorithm (represented as GMOPSO). The algorithm combines the methods of mutation operator, annealing factor and feedback mechanism to improve the diversity of the population on the basis of meeting the stable motion, avoiding the local optimal solution and accelerating the convergence speed. By adopting the average optimal evaluation method, the robot arm motion trajectory has been testified to constructively fulfill the picking standards of stability, efficiency and lossless. The performance of the algorithm is verified by ZDT1~ZDT3 benchmark functions, and its competitive advantages and disadvantages with other multi-objective evolutionary algorithms are further elaborated. In this paper, the algorithm is simulated and verified by practical experiments with the optimization objectives of time, energy consumption and pulsation. The simulation results show that the solution set of the algorithm is close to the real Pareto frontier. The optimal solution obtained by the average optimal evaluation method is as follows: the time is 34.20 s, the energy consumption is 61.89 °/S2 and the pulsation is 72.18 °/S3. The actual test results show that the trajectory can effectively complete fruit picking, the average picking time is 25.5 s, and the success rate is 96.67%. The experimental results show that the trajectory of the manipulator obtained by GMOPSO algorithm can make the manipulator run smoothly and facilitates efficient, stable and nondestructive picking.

When picking strawberries outdoors, due to factors such as light changes, obstacle occlusion, and small target detection objects, the phenomena of poor strawberry recognition accuracy and low recognition rate are caused. An improved YOLOv5n strawberry high-precision recognition algorithm is proposed. The algorithm uses FasterNet to replace the original YOLOv5n backbone network and improves the detection rate. The MobileViT attention mechanism module is added to improve the feature extraction ability of small target objects so that the model has higher detection accuracy and smaller module sizes. The CBAM hybrid attention module and C2f module are introduced to improve the feature expression ability of the neural network, enrich the gradient flow information, and improve the performance and accuracy of the model. The SPPELAN module is added as well to improve the model’s detection efficiency for small objects. The experimental results show that the detection accuracy of the improved model is 98.94%, the recall rate is 99.12%, the model volume is 53.22 MB, and the mAP value is 99.43%. Compared with the original YOLOv5n, the detection accuracy increased by 14.68%, and the recall rate increased by 11.37%. This technology has effectively accomplished the accurate detection and identification of strawberries under complex outdoor conditions and provided a theoretical basis for accurate outdoor identification and precise picking technology.