Impact Analysis of CNN-Based Machine Vision and Soft Robotic Grippers on Labor Efficiency in Automated Berry Harvesting

Berry harvesting is one of the most labor-intensive activities in agriculture due to the delicate structure of the fruit, clustered growth, and strict quality requirements. Increasing labor costs and seasonal shortages further strain manual harvesting, threatening farm productivity. Recent advances in robotics integrate convolutional neural networks (CNNs) for accurate fruit detection and soft robotic grippers for delicate handling. This study analyzes the impact of these technologies on labor efficiency through a desk-based review of published data. Metrics such as detection accuracy, picking cycle time, fruit damage rates, and estimated labor substitution were extracted from peer-reviewed research and technical reports. Results indicate that CNN-based systems achieve fruit detection accuracies above 90%, while soft grippers maintain damage rates below 5%. Harvesting robots achieve cycle times of 4–10 seconds per berry, allowing substitution of approximately two to five human pickers. However, limitations remain, including high costs, technical complexity, and reduced performance under variable field conditions. These findings highlight the potential of CNN-based vision and soft robotic grippers to improve labor efficiency in berry production, while also emphasizing the need for further innovations and cost-effective adoption models.