IMPROVING ACTIVE ACCIDENT PREVENTION IN ROBOTIZED MANUFACTURING PLANTS

This work describes the design strategy and preliminary testing of an active protection system characterized by low-cost commercial sensors of proven reliability. In the system, each risk level is defined and a particular robot reaction assigned to it. The risk calculation is bas ed on the recursive solution of the manipulator's kinematic problem fo r the evaluation of the horizontal distance components and the relativ e speed between operator and machine. Computational effort is minimize d, since (1) the protection system procedures can be performed directl y in the robot controller with the robot kinematic and dynamic calcula tions and (2) only the cylinder-shaped volumes around the operator's e stimated position have to be defined in order to perform a probabilist ic evaluation of the risk levels. Preliminary testing on a simplified setup has validated the system's reliability, its insensitivity to bac kground disturbances, and the ease of use of the sensors.