Traditional agricultural harvesting systems are facing increasing challenges, most notably high labor costs of more than 30% in some countries, lack of qualified labor, and mechanical losses that exceed 25% in sensitive crops such as tomatoes and strawberries.

In contrast, the last three decades have seen a remarkable development in the integration of AI into harvesting techniques. Multiple studies have shown that computer vision algorithms such as YOLO and Faster R-CNN have increased fruit identification accuracy to more than 90%, while convolutional neural networks have helped reduce early and late harvest rates by between 25% and 35%.

This review was based on the PRISMA methodology and covered a systematic analysis of 55 studies published between 2000 and 2025, covering diverse crops, open and protected agricultural environments, tracking technologies, algorithms, applications, and challenges.

The analysis showed that 62% of the studies focused on high-value crops (cherries and tomatoes), and that intelligent robots reduced the need for labor by up to 70%, but remained expensive with an average cost between $35,000 and $120,000.