Safe and flexible human-robot cooperation in industrial applications

The increasing need of a closer cooperation between human and robots in industrial environments requires the appropriate treatment of risky situations that may arise in the shared working space. This paper deals with technological approaches about the prediction of such situations and the avoidance of the hazards, by using a flexible monitoring system that involves the fusion of information and data from various sources: a set of stereo cameras is used to build a 3D representation of the scene, which builds the basis for the danger estimation. Representative features of the elements of the scene, such as the position of the robots and humans, their trajectories and relative velocities are used to predict future hazardous situations, which are resolved by an appropriate reaction on the behavior pattern of the robots. Additionally, the authors present a visualization tool that enables the pre-selection of the most suitable configuration of the system according to the needs and characteristics of particular use cases, as well as the definition of the danger situations and the corresponding reactions.