PEARL - New technology for magnetic fixation of hides in frame drying processes
Feltre is an Italian company that invents, designs and manufactures automation for tanneries. The company offers a complete range of standard and special stackers, barrel loading and unloading systems, sorting lines, transfer systems for all types of leather and various solutions for leather handling and manipulation, as well as a complete range of machinery for raw hides and special machines.
The company needs an hw and sw architecture analysis for a leather boundary recognition system to automate the positioning of magnetic bodies using a robotic system. Feltre decided to hire the Artificial Intelligence Research and Innovation Center (AIRI) for the task. From this idea comes the collaboration between AImageLab and Feltre, resulting in the research programme entitled “PEARL - New technology for magnetic fixation of hides in frame drying processes”.
The project consists in the definition of an Artificial Intelligence and Machine Vision solution for a System for automatic fastening by positioning magnetic spheres over wet hides laid on perforated metal mesh.
The following hypotheses will be discussed in this framework:
Hypothesis 1: A camera positioned centrally above the net (Net size approximately 3500X3000mm), with Hardware Definition needed to detect edges of the skin lying on a fixed net relative to the camera.
Hypothesis 2: Linear cameras positioned at the skin inlet. Hardware definition needed to detect skin edges lying on a net moving perpendicular to the cameras.
It is expected to study and develop a prototype software, which will be delivered in python language for a general-purpose computer, possibly equipped with GPU accelerators. The software will have the task of:
1) Recognition of the x,y coordinates of the leather edges with respect to a reference point on the network
2) Intersection of the coordinates of the leather edges with the positions of the translators for depositing metal balls
- determination of the x-coordinate on the leather edge for each sideshifter whose y-position is known
- determination of the y coordinate on the leather edge for each sideshifter whose x position is known
3) Optimisation of sphere positioning coordinates according to the mesh hole that is innermost on the edge
- determination of offset x coordinate with respect to the mesh hole innermost on the edge for each sideshifter whose y position is known
- determination of the y-coordinate offset with respect to the mesh hole innermost on the edge for each Sideshifter whose x position is known.
The algorithm and the software that derives from it will have as input various project parameters (such as the dimensions of the X, Y sides of the net, the diameter of the holes in the net and their pitch with respect to the point of origin, the diameter of the magnetic spheres, the y coordinates of the translators positioned on the short side of the net, the x coordinates of the translators positioned on the long side of the net, any offset of the net origin position with respect to the coordinate system of the translators).
Input of the algorithm and of the prototype software will have as input the Image of the skin detected by camera and will have as output the data such as the coordinates x of the translators.
Output the data such as the x coordinate closest to the net hole more internal to the edge for each translator whose position y is known, the y coordinate closest to the net hole more internal to the edge for each translator whose position x is known.