ASHS 2025 Conference

Strawberry harvesting is a labor-intensive task that requires careful, selective picking to avoid damaging delicate fruit. To address labor shortages and enhance harvesting efficiency, we propose an AI-enhanced soft robotic system capable of autonomous and selective strawberry harvesting. Specifically, the system incorporates advanced computer vision algorithms, leveraging the Grounding DINO model, to detect and localize ripe strawberries with high accuracy. A compliant soft robotic gripper, guided by real-time perception, then gently harvests only the target fruits, minimizing potential damage to adjacent berries and plants. Experimental results demonstrate that the system achieves a ripe strawberry detection accuracy of 88% and a harvesting success rate of 66.67%. This integrated approach offers a scalable, precise, and labor-efficient solution for modern strawberry production.