MYCOLab Ai Logo - gold and army green

The Application of an Intelligent Agaricus bisporus-Harvesting Device Based on FES-YOLOv5s

Summary

Researchers developed an intelligent robot that automatically harvests button mushrooms with high precision. The system uses AI-powered camera vision to identify mature mushrooms in crowded growing beds, then carefully picks them with a gentle robotic arm. Testing showed the robot successfully harvests over 94% of mushrooms while causing minimal damage, making commercial mushroom farming more efficient and cost-effective.

Background

Agaricus bisporus (button mushroom) is a nutritious, widely consumed food crop with high economic value. However, mushroom harvesting remains largely manual, resulting in low efficiency, high damage rates, and poor quality, hindering industrial development. Automated harvesting solutions are needed to improve productivity and reduce costs.

Objective

To design and develop an intelligent, machine vision-based harvesting device for Agaricus bisporus that operates efficiently in multilayered cultivation spaces while minimizing crop damage. The device combines the FES-YOLOv5s deep learning model for accurate mushroom detection with S-curve motor control algorithms for stable, precise robotic harvesting.

Results

Recognition accuracy averaged 96.72% with 2.13% missed detection rate and 1.72% false detection rate. Harvesting success rate was 94.95% with 2.67% average damage rate and 87.38% harvesting yield rate. S-curve motor control demonstrated superior vibration reduction compared to trapezoidal control, with lower maximum vibration velocity and displacement across all three axes.

Conclusion

The intelligent harvesting device successfully meets requirements for automated Agaricus bisporus harvesting with high accuracy and low damage rates. The FES-YOLOv5s model effectively handles dense, overlapping mushrooms in dim growing environments. These results advance development of intelligent harvesting robots for industrial mushroom production.
  • Published in:Sensors (Basel),
  • Study Type:Applied Research/Engineering Development,
  • Source: 10.3390/s25020519; PMID: 39860890
Scroll to Top