Our technology
Aigritec uses state-of-the-art technology to create our solutions. Below you find an overview of our current developments.
1 Advanced robot components
Aigritec uses latest generation robot components that are highly modular and that guarantee high speed, precision and resistance.
2 Robots powered by Deep Learning
Aigritec uses state-of-the-art Deep Learning to automate vision, motion control, multi-arm planning, motor tuning, and much more!
3 Computer Vision
We detect fruits by running object detection using state-of-the-art computer vision models. In this video, we show inference of our RetinaNet model. We use an end-to-end pipeline to finetune this RetinaNet model on any data that we gather. Hence, we can easily scale up to different fruits and different environments. The RetinaNet detects the fruits it has encountered during training and defines bounding boxes around these fruits in real time.
4 Fruit localization
In order to automate the harvesting procedure we extract real world coordinates from the fruits that have been detected by our computer vision model. We use a RealSense d435i depth camera to correctly measure the depth and therefore distance of the fruit, and calculate its position with respect to the arm. Once the location has been precisely determined, we simply move the arm to that location and instantiate the picking procedure.