Safety Issues for Human-Robot Copperation in Manufacturing Systems

Agostino De Santis, Bruno Siciliano

Tools and Perspectives in Virtual Manufacturing, July 10^th 2008 – July 11^th 2008

Summary

                  The next generation robots will be enhanced with the challenging and revolutionary feature of physical Human Robot Interaction (pHRI), making safety and dependability key concepts of the future of robotics.

Safety, being fundamental for future HRI, appears to cover several aspect, such as: mechanics, electronics and software. Up to now safety has force robots to be segregated from the operators working environment, cHRI (cognitive Human Robot Interaction) has been commonly debated in the scientific community, however robots are distinct from computers and other machines: they generate force and have “body”, making it be an intelligent connection in the direction of pHRI, allowing more and more safe, fast and accurate motions without third-party sensors.

The new robots are then considered under two criteria: safety (which includes also “mental safety”, which is the awareness of robot motion) and dependability (which allows “human-in-the-loop” conditions). Standards at the moment are not yet ready for the share of operational space, work is being going on the international ISO 10218 in order to include aspect on robot’s work place.

It is clear that in order to allow humans and robots to share common working environment, metrics regarding safety levels must be introduce and not surprisingly measures already use in the automotive sector are currently used. Two methods are applied: direct interaction (for head collision with another solid object) and indirect interaction (for sudden head motion with no direct contact). The scaling used (AIS – Abbreviated Injury Scale) is indicating the injury severity on the overall injury (MAIS); injury types are then divided in a classification related to type and consequences and is based on an ascending scale from 0 to 6. Paradoxically the result from this metric is that it enourages the usa of robots, since it is considered that even withouth the use of robots operator get injured. Safety can be taken with two different approaches: intrinsic safety and actuation (for example a distribuition macromini actuation DM² [Zinn,2002], where for each d.o.f. a pair of actuators connected in parallel and located in different parts on the manipulator, ensuring lower inertia and good performance; VIA – Variable Impendance Approach, is a mechanical control co-design that allows varying stiffness, damping and gear-ratio in order to minimize the negative effects of control performance) and safety by mean of control (which could be either position controlled or force/impendance control, the latter being direct in the case of a feedback loop control or indirect with a motion control loop). In the case of safety by mean of control, we my consider reactive control obtained through potential fields, which have the mission of creating attracting or repelling volumes, an example is the skeleton algorithm, which creates rotational volumes in proximity of links, this created a virtual region which approaches the real volume of a considered part of a manipulator.

The algorithm allows to use the human head avoidance illustrated previously and in case of a safe robot, the additional safety can be considered to be enough.

The issues which the authors encounted are regarding communication, in fact a emergency stop of the robot still has to be tested in case of particular necessities, but still it appears to be a fast modelling method.

The importance of simulation useing virtual reality is underlined, since it provides and ergonomic evaluation, comfort measure and of course a simulation of possible and eventual malfuctioning, allowing a fast comparison of interface, appearance and kinematic parameters.

Key Concepts

Safety, Virtual Reality, Simulation, Reactive Control, skeleton algorithm, physical Human Robot Interaction.