Developing a functional model of a two-camera intelligent safety system for detecting dangerous situations at a robotized workstation

The project involved constructing a neural system that made it possible to determine the spatial location of an object entering a zone monitored with cameras. A PC-compatible program implementation of the developed neural system was also prepared. Specialist computer software for generating synthetic stereoscopic images was also developed.A vision system consisting of two professional matrix cameras was developed. Procedures for calibrating cameras and for preparing a reference depth map were also prepared. Then, the vision system was integrated with the new neural system. The last part of the project consisted in testing the developed neural system in virtual reality, with a functional model of the vision safety system. Synthetic images made it possible to use the computer model to test an assembly line and to compare the operation of a light curtain with a vision safety system.