Modeling and assessing an intelligent system for safety in human-robot collaboration using deep and machine learning techniques

The introduction of technological innovations is essential for accident mitigation in work environments. In a human-robot collaboration scenario, the current number of accidents raises a safety problem that must be dealt. This work proposes an intelligent system that aims to address such problems using deep and machine learning techniques. More specifically, this solution is divided into two modules: (i) collision detection between humans and robots and (ii) worker's clothing detection. We evaluated these modules separately and concluded that the proposed intelligent system is efficient in supporting safe human-robot collaboration. The results achieved a sensitivity level greater than 90% in identifying collisions and an accuracy above 94% in identifying the worker's clothing.