AI-Powered Strawberry Harvester Using a Robotic Arm.

Understanding how to integrate robotics and artificial intelligence is essential for advancing automation in the agricultural sector. However, converting AI outputs into precise mechanical movements of robotic arms remains a significant challenge. We investigate how to combine AI-based computer vision with real-time robotic motion control. Using an Intel RealSense D435 depth camera, an NVIDIA Jetson Orin platform running a pretrained AI model, and a MyCobot 320 Pi robotic arm, we implement inverse kinematics to guide the arm toward targets. Our results show a positional root-mean-square error (RMSE) of 8.0 mm, and GPU-accelerated processing improves AI inference speed by nearly two orders of magnitude. These findings provide insight into automating harvesting tasks and suggest strategies for enhancing agricultural efficiency through intelligent robotics.