Technologies
9 October 2024
ROS system for robot movement
Have you ever wondered how robots manage to move intelligently, making their own decisions? Let's delve into this robotic capability by exploring ROS.
A few weeks ago you were able to learn about this project on our RRSS and today we bring you an entry dedicated to it. You can also find out more details on its own website.
This project arises from the need, existing in light packaging waste sorting plants, to make it easier for operators to remove bulky waste from the treatment line that, due to its shape or size, is detrimental to subsequent sorting processes.
To do this, we are creating an artificial intelligence system for the identification and classification of waste through artificial vision, whose initials are SIARA. We tell you!
What is SIARA?
It is a computer vision system based on deep learning (neural networks) for the detection and separation of bulky waste found on the conveyor belt at the entrance to the plant for the selection of plastic waste, cartons or cans (from the yellow container) and thus prevent the line from clogging or being damaged by such waste.
With its design different problems will be solved:
- Reduce process time
- Minimize costs
- Avoid tedious tasks
It is a project financed by the Ministry of Economy and Business and co-financed by the FEDER or European Regional Development Fund, the objective of which is to strengthen socio-economic cohesion within the European Union by correcting imbalances between its regions.
How does SIARA work?
The system will learn to detect the waste that does not have to be removed
and to dispose of the rest through cameras with different sensors and an artificial intelligence system. As we have said, deep learning will be used to train SIARA, creating a database that will include the acquisition of images and their tagging, a very important work that will be carried out during the first stages of the project.
Most computer vision applications use cameras that capture the visible spectrum, colors, and textures, making it easier to distinguish plastics, cartons, and cans that may have the same color or texture. For further optimization, the use of multispectral cameras will be studied. Likewise, 3D cameras provide depth and can help select debris areas on the conveyor belt in the image.
It is also necessary to choose the sensors to include, the computer and the lighting and the position of all the elements in the structure, as well as the hardware that will be used for their evaluation.
En ella, una cinta transportadora recibe los residuos de entrada y los distribuye a lo largo de la planta en distintos procesos de selección. El sistema de adquisición que se va a utilizar para la captura de datos se instalará sobre esta cinta de alimentación de la línea y tomará imágenes de los residuos que entren a la planta. Este sistema se seleccionará y se calibrará en los laboratorios de ATRIA.
Where is SIARA applied?
The test environment of the project will be the plant for the selection of plastic containers, cans and briks titled by the Mancomunidad de San Marcos, in which Trienekens operates, in Urnieta (San Sebastián).
In it, a conveyor belt receives waste and distributes it throughout the plant in different selection processes. The acquisition system that is going to be used for data capture will be installed in this line on the feeding belt and will take images of the waste entering the plant. This system will be selected and calibrated in the ATRIA laboratories.
If you are interested in learning more about our project, do not hesitate to visit the Project’s own website www.siaraproject.es.
In addition, you can follow us on Instagram to find out about the latest news and the development of SIARA until its final phase and launch. We really want to tell you every day!
If you want to ask us about this classification system based on deep learning, contact us!
