Artificial intelligence robot learns to pick tomatoes

Нехватка рабочей силы в сельском хозяйстве подталкивает его к большей автоматизации, особенно в части сбора урожая. But not all crops are easy to machine. For example, tomatoes grow in clusters, which means that the robot must carefully select ripe fruits, leaving unripe ones untouched, writes xrust. This requires precise management and informed decision making.

To solve this problem, Associate Professor Takuya Fujinaga of Osaka Metropolitan University's Graduate School of Engineering has developed a system that trains robots to judge how easy each tomato is to pick before attempting to pick it.

His approach combines image recognition with statistical analysis to determine the optimal angle to harvest each fruit, explains sciencedaily. The robot analyzes visual details such as the tomato itself, its stems, and whether it is hidden behind leaves or other parts of the plant. This data helps the robot choose the most efficient way to approach and pick up the fruit.

From detection to decision making

This method departs from traditional systems that only focus on detecting and identifying fruits. Instead, Fujinaga introduces what he calls an “ease of harvesting score.” «This goes beyond just asking 'can a robot pick tomatoes?' and moves on to thinking about ‘can a robot pick tomatoes like a human?’, which is more relevant to real-life agriculture,” he explained.

During testing, the system showed 81% success rate, exceeding expectations. About a quarter of the successful harvests were of tomatoes picked from the side after the first attempt at picking from the front failed. This suggests that the robot can adjust its approach if the first attempt fails.

The study highlights how many variables affect robotic harvesting, including how tomatoes cluster in clusters, the shape and position of stems, surrounding leaves and visual obstructions. “This research establishes 'ease of harvesting' as a quantifiable metric, bringing us one step closer to creating agricultural robots that can make informed decisions and act intelligently,” Fujinaga said.

The future of robots and people in agriculture

Looking to the future, Fujinaga envisions robots that can independently determine when crops are ready to be harvested. “This is expected to usher in a new form of agriculture where robots and humans collaborate,” he explained. “Robots will automatically pick the easy-to-pick tomatoes, while humans will handle the more difficult-to-pick fruits.”

Xrust Artificial intelligence robot learns to pick tomatoes