Vegebot uses machine learning to harvest crop

Jaehyun Park

The vegebot lettuce-picking robot(University of Cambridge)

Everyone knows technology is always evolving and by no means is agricultural implement an exception.

In these days, a research team at Cambridge University has developed a lettuce harvesting robot ‘Vegebot’ which is based on a machine learning. Also the team think the technology can be used to other crops in order to solve the shortages of agricultural labor and cut food waste.

Agricultural harvesting is a part of ​​industry that has always accepted new technological advances. Crops like wheat or potatoes are relatively easy to mechanically harvest on a large scale, but other crops require labor that is difficult to harvest. For example, developing an autonomous robot system that picks out vegetables like cucumbers is a huge challenge.

Lettuce is another difficult crop that has proved so far resistant to automation. Not only does it need to grow flat on the ground and cut at the right spot, the field of lettuce presents a new challenge to robot vision, which must be at home on the individual heads of lettuce ready to be harvested from the sea of ​​green leaves.

Photo by Franck V. on Unsplash

However, The resulting machine, called Vegebot, showed impressive identification with 91% localization success. This system still needs work before being converted into a commercial solution, especially high damage rates of about 38% are reported. The researchers note that this rate of damage means that a significant amount of the entrapped variety did not meet supermarket standards, but still remains edible.

Another obstacle to overcome is the speed of the system. Vegebot takes about 32 seconds to pick each lettuce. This is much slower than the average human picker, but researchers suggest that lighter production materials should accelerate this process.

In a positive side, robots can potentially reduce food waste only if they perform multiple passes in a single area and only pick specific fruits and vegetables. Current human harvesting methods generally sweep the field at once, picking unripe fruits or untreated fruits and discarding them later.

Source: University of Cambridge, Journal of Field Robotics.

read original article at——artificial_intelligence-5

Do NOT follow this link or you will be banned from the site!