News

Our nervous system does an amazing job of tracking sensory information, and it does so using signals that would drive many computer scientists crazy: a stream of activity information that can be transmitted to hundreds of additional neurons, which are integrated with similar streams of information coming from other neurons.

So now, researchers are using information circuits to build an artificial robotic skin, adopting some of the principles of how signals from our sensory neurons are transmitted and integrated.

The system uses some non-neural features, but it has the advantage of giving us chips that can run neural networks using information signals, which would allow this system to be integrated with some energy-efficient hardware running artificial intelligence control software .

There are four ways in which these signal trains can convey information: by the shape of a single pulse, by their magnitude, by the length of the information signal, and by the frequency of the information signal. The frequency of the information signal is the most commonly used means of transport in biological systems, and researchers use it to convey the pressure experienced by a sensor.

The remaining information is used to create a barcode that helps identify the sensor that the reading came from. In addition to recording pressure, the researchers had each sensor send an “I’m still here” signal at regular intervals. Not receiving this signal would be an indication that something has gone wrong with that sensor.

The information signals allow the next level of the system to detect any pressure being applied to the skin, as well as where it came from. This level can also make basic assessments of sensory input:

“Raw pulses initiated by pressure from the pulse generator accumulate in the signal buffer center until a predetermined pain threshold is exceeded, triggering a pain signal.”

This can enable basic reflex responses that don't involve higher-level control systems. For example, the researchers set up a robotic arm covered in their artificial skin and made it move whenever it experienced pressure that could cause damage.

The second layer combines and filters signals from the skin before sending the information to the arm controller, which is the equivalent of the brain in this case. So the same system can make a robotic face change expressions based on the pressure its hand “feels.”

The artificial skin is designed to be assembled from sections that can snap together using magnetic locks. These automatically connect any necessary wiring, and each section of skin can transmit a unique identification code.

So if the system detects damage, it is relatively easy for an operator to remove the damaged part and replace it with new material. The researchers call their development neuromorphic robotic electronic skin, or NRE-skin, from neuromorphic robotic e-skin.

The term “neuromorphic” is a bit vague, with some using it to indicate a technology that follows the principles used by the nervous system.

DOI: https://doi.org/10.1073/pnas.2520922122

0 reviews

iGuRu.gr

Join the discussion

Your email address will not be published. Required fields are marked *

Your message will not be published if:

  1. Contains abusive, defamatory, racist, offensive or inappropriate comments.
  2. It causes harm to minors.
  3. It harasses the privacy and individual and social rights of other users.
  4. It advertises products or services or websites.
  5. Contains personal information (address, telephone number, etc.).