Vision-Based Fault Diagnosis and Self-Recovery for Strawberry Harvesting Robots Published

A paper led by PhD student Meili Sun has been published in Artificial Intelligence in Agriculture, titled “Vision-based early fault diagnosis and self-recovery for strawberry harvesting robots”.

The study presents an integrated vision framework that enables early fault diagnosis and self-recovery during robotic harvesting. Key innovations include:

• SRR-Net: An end-to-end multi-task network for unified fruit and gripper perception.
• Relative error compensation: Reduces picking point misalignment from 11.50 mm to 3.12 mm and 5.25 mm to 4.06 mm along the x- and y-axes, respectively.
• Early abort strategy: Detects empty grasp/misgrasp early, saving approximately 0.5 s per failure case.
• Slippage prediction & recovery: Achieved 88.89% prediction success and 81.25% recovery rate for slipping strawberries, saving 4.00 s per failure cycle.

A video demonstration is available at https://youtu.be/UOfwlHgXUgU.

DOI ↗

Congratulations to the authors! 🎉

我实验室博士生孙美丽为第一作者的论文 《基于视觉的草莓采摘机器人早期故障诊断与自恢复方法》 在 Artificial Intelligence in Agriculture 期刊正式发表。

该研究提出了一套集成视觉感知、故障诊断与自主恢复的完整框架，主要创新包括：

• SRR-Net端到端多任务网络：统一感知果实与夹爪状态
• 相对误差补偿：将采摘点定位误差从11.50 mm和5.25 mm分别降至3.12 mm和4.06 mm
• 早期中止策略：提前检测空抓/误抓，每次失败节省约0.5 s
• 滑移预测与恢复：滑移预测成功率88.89%，恢复成功率81.25%，每次失败节省4.00 s

视频演示请见：https://youtu.be/UOfwlHgXUgU

DOI ↗

祝贺论文作者团队！ 🎉