This project aims at developing a demonstrator that integrates state-of-the-art technologies and improve it on a use-case representative of the industrial world.
The demonstrator will consist in a robotic / cobotic arm coupled to an acquisition sensor (RGBD type). This device will be positioned in a workspace made of a rack / shelf containing objects / pieces of various shapes and qualities (materials, densities, colors ...) in front of which will be placed a typical conveyor prototype of industrial installations. The type of tasks expected to be carried out by the demonstrator will be "pick and place" type tasks where an object will have to be identified in shelf and then placed on the conveyor.
This type of demonstrator will be closer to the real industrial conditions of use than the "toy" examples used in the academic field.
This demonstrator will focus first on the short-term effectiveness based on state of the art technologies for both hardware and software, for a use case representative of the industrial world.
At first, it will thus be less focused on the evolution of the algorithms used than on the adaptation of the parameters, the injection of knowledge a priori dependent on the context making it possible to reduce the high-dimensional input space, etc.