Artificial Fingerprints Help Robots Distinguish Shapes

In recent years, biomechanical engineers have discovered that fingerprints don’t just improve grip, they also perform a delicate type of signal processing. The ridges modify vibrations in the skin in a way that best allows human nerve endings to receive them and this allows the skin to extract much more information from touch than it otherwise would. We discussed this idea almost two years ago.

Having discovered the effect in humans, there was a certain inevitability to the idea that researchers would apply the same technique to robot touch. Today, Saba Salehi, John-John Cabibihan and Shuzhi Sam Ge at the National University of Singapore end the waiting.

These guys have developed a robotic touch sensor, covered it with fingerprints and worked out exactly how useful they are. Their sensor consists of four force-sensitive sensors on a 4mm square plate. They covered this plate with a flat plastic sheet and then a ridged plastic sheet and recorded the change in force measurements as the plate touched different objects.

The results are surprising. The ridged fingerprints don’t just help with grip and texture, they can also identify the shape of the object being touched. Salehi and co say that the pattern of forces that the fingerprint generates can help distinguish between flat and rounded objects.

That should turn out to be useful. It’s part of a growing movement in robotics to outsource information processing to the very shape, structure and material properties of the robot. So-called morphological computing can dramatically reduce the processing load that a central processor has to take on and seems to be how nature has solved the problem for living things.

It also means that human fingerprints may play a similar role, helping us not only with grip and texture but also with shape recognition. Anybody know of any existing evidence that this is the case?

Ref: arxiv.org/abs/1109.3688: Artificial Skin Ridges Enhance Local Tactile Shape Discrimination