Design of Robotic Operation System for Fully Autonomous Inspection of Strawberry Pests and Diseases in Facility Ridged Cultivation

Abstract

In response to the frequent occurrence of strawberry pests and diseases, along with the labor-intensive and time-consuming nature of regular inspections, this paper presents a comprehensive design scheme for a fully autonomous single-plant sampling strawberry pest and disease inspection robot. Key components such as the canopy's curved multi-viewpoint image localization acquisition system and the motorized trans-monopoly self-propelled precise positioning chassis are meticulously designed. Moreover, a fully autonomous perception and control system for the robot is developed, leading to the completion of the system development phase. Additionally, the trained pest and disease detection model is seamlessly integrated into the robot's Jetson Nano through the API interface. Subsequently, the robot undergoes rigorous field testing for the inspection of multiple classes of strawberry pests and diseases, the detection accuracy, false detection rate and leakage rate of the robot for pests and diseases at manually selected canopy locations were 92.10%, 6.58% and 1.32%, respectively. Finally, the robot successfully fulfills the intended design objectives, marking a significant stride towards addressing the autonomous detection needs in strawberry production.