TUM creates seal of approval for robotics
When autonomous robots interact with humans and perform tasks independently, their sensitivity is crucial. This is what makes them safe and flexible. However, there is currently no standardised procedure for determining this sensitivity and thus individual robots’ fitness for physical interactions. Researchers at the Munich Institute of Robotics and Machine Intelligence (MIRMI) at the Technical University of Munich (TUM) have now developed an evaluation scheme that can be used to compare industrial robots and, in the future, other systems such as mobile and humanoid robots or robotic hands.
Achim Lilienthal, Deputy Director of MIRMI and Professor of Intelligent Systems Perception, is convinced: “The TUM MIRMI test methodology has the potential to become an industrial testing standard.” This would create the conditions for establishing a seal of quality for the fast-growing robotics market. “Knowing the performance capabilities of a robot system is an enormous help to industry, where companies want to make the best possible use of robotic systems.”
TUM MIRMI Executive Director Prof. Lorenzo Masia says: “Based on the original idea, I am certain that the AI Robot Safety and Performance Centre at TUM MIRMI will develop into an independent national testing centre for robotics.”
Single-armed robots as a starting point
As a first step, the researchers examined and categorised single-armed robots from well-known manufacturers that are used in industry and research. Although many robotic arms look similar, they each have their own strengths and weaknesses, with big differences between sensors, motors and the actual brain of the intelligent machines. This results in systems that differ greatly in their basic capabilities. Some are powerful and precise in their movements, while others are gentle, flexible and sensitive in their interaction with their surroundings.
To highlight these differences and the continuous progress of robotics, researchers at TUM MIRMI’s AI Robot Safety and Performance Centre have developed the Tree of Robots. Based on Charles Darwin’s Tree of Life, it depicts the fundamentally different adaptations of various species to their ‘habitat’, in this case not of living beings, but robots. “To do this, we look at the basic skills a robot needs in processes, such as how well it follows a given path, takes up a position, how gently it can make contact with surfaces, and how safe it is in dealing with collisions between robots and humans,” explains lab director Robin Kirschner.
25 measurements of robot sensitivity
For tactility alone, there are 25 measurements that describe a robot’s sensitivity in physical contact with its surroundings. Among other things, this involves finding out whether the force applied to a surface, for example, is not actually higher than intended or whether a robot could cause injury to humans. From the pattern that emerges on a spider diagram based on the 25 measurements, even laypeople can see the robot’s sensitivity at a glance.
A clear picture of strengths and weaknesses
Based on the performance of individual systems, researchers categorise them as ‘industrial robots,’ ‘cobots,’ ‘soft robots’ and ‘tactile robots.’ For surgical robots the decisive requirement is precision. For warehouse or factory applications, by contrast, the emphasis is on strength and resilience, i.e. the ability to perform repetitive movements over long periods. “We combine existing motion metrics with our new tactile metrics to provide the first-ever overview of the totality of the basic capabilities for physical interactions of a robotic system,” explains Kirschner.
