Machine Learning Software for Detecting Ripe Tomatoes using Artificial intelligence

The Danish Technological Institute, together with the Funen horticulture Alfred Petersen & Søn, robot integrator Egatec A/S and Belgian Bogaerts have developed the world’s first intelligent tomato-picking machine. Sun-ripened tomatoes in Denmark are hand-picked manually every day since there are no robots that are programmed to handle the task. 

Tomatoes are plants that must be continuously defoliated and tied up to get sunlight and then it must be assessed whether they are ripe enough to be harvested. There are currently no ways to determine and differentiate a ripen tomato by machine algorithm therefore, the harvest of the plant is still by hand.

In the video, the machine is seen in the training process of a deep learning algorithm to detect ripe tomatoes using a 3D camera. It aims to assess the ripeness of the tomatoes before cutting the clusters. The tomato robot is able to and locate the tomato clusters and move around the greenhouse with sensors and camera technology to collect data. 

The robot learns by taking thousands of pictures, which is then put into a system from which the algorithm must assess whether the tomato is ripe enough for harvesting. The Danish Technological Institute developed an advanced 3D camera and brain with an artificial intelligence algorithm learns to recognize different degrees of maturity in tomatoes through machine learning.