Robot Getting Trained to Feel and React to Pain

One great fact about robots is that they do not feel pain and for this reason they are most preferred to put on to dangerous work or perform tasks that can be fatal to humans. But as per a team of German researchers, feeling pain can be a good skill for robots to possess.

 

The researchers from the Leibniz University of Hannover are experimenting to develop an ‘artificial nervous system for robot’ to teach them the sensation of pain and deliver a quick reaction against it. Such a development aims to protect them experiencing potential damage to their gears, electronics, and motors. The declaration about the project was made at the International Conference on Robotics and Automation (ICRA) that was held in Stockholm, Sweden.

 

Feeling pain is a good idea for robots. Johannes Kuehn states the reason for same as, “Pain is an element that protects us as when we evade from the pain source then it helps us to avoid getting hurt. Humans do not have the capability to get injured and feel pain more often because their bodies instantly react against things that hurt them.”

 

 

 

Kuehn worked on this project with Professor Sami Haddadin, a famous expert in physical human-robot interaction and safety. Both of them agrees with the ideology that if robots will be able to sense pain and will try to protect themselves, then they will also understand the painful conditions of humans and will also try to protect them. Their reasoning is based on the biological mechanism of humans to sense and react to pain.

 

This idea of inculcating a nervous system inside the robots is based on the reflex controller, which was previously used by a team of Stanford and University of Rome to assist a robot arm in avoiding colliding with people. Kuehn and Haddadin mentioned in the ICRA paper that a robot must be able to identify and differentiate unforeseen physical disturbances and states, rate the possible damage they may encounter with and instinctively take countermeasures, known as reflexes. To tackle the demanding requisite, this human antitype will serve as an inspiration implying that human pain-reflex actions are utilized for designing according to robot pain reaction and controls.

 

The robot nervous tissue model is inspired by the skin structure of human to evaluate the level of pain that they must feel for a specific amount of force. This model transfers the pain information in repetitive spikes to the body, just like the way it is transferred in the human neurons. If the force exceeds a specific level, then the pain controller will transmit a signal classifying the level of pain into moderate, light or severe.

 

Conclusion – regarding control and bio-inspiration, the paper by Kuehn and Haddadin is just the first step towards a reflex controller against the pain for robots. While it is expected that it would bring great utility to the robots, it is also an undeniable fact that the robots must be trained to not only safeguard humans but also to save themselves from getting hurt.