Everyone knows that robots don’t feel pain. If you tell a robotic arm that it’s fat, it won’t break down in tears. Pain is a feeling reserved for living, breathing, biological organisms with a nervous system. People can feel pain, animals can feel pain, but robots cannot feel pain. Even animal-shaped robots can’t feel pain. Kick a robotic dog, and it will maintain its balance and continue about its business as if nothing happened.
Robots don’t feel pain, but a team of researchers is trying to develop an artificial nervous system that would allow robots to sense, classify, and rate the potential for damage of “unforeseen physical states and disturbances”.
It might be an exaggeration to say that this would allow robots to feel pain, but the researchers are using how humans respond to pain to in order to avoid injury as their model.
The two German researchers are trying to develop a robotic nervous system that would allow a robot to sense “pain”, and use that information to avoid damage to itself, its environment, or people nearby.
Robots with this artificial nervous system would still remain unfazed by jeers about their weight, but they would be able to detect physical “pain” and respond accordingly to prevent injury or damage.
Pain helps humans avoid injury. If we touch something that’s hot or sharp, we pull our hand back so that we don’t get hurt. The idea is to give robots the ability to sense and classify a physical presence by touch in similar way.
But should robots be able to feel pain?
The fact that they don’t feel pain is currently one of the big advantages of robots. Robots can be used in environments that would certainly be painful – and often lethal – for humans.
But the researchers believe that it’s important for robots to feel pain for the same reason it’s important for humans to feel physical pain – it protects us from harm. Once we start to feel pain, we can adjust to prevent injury.
Robots would be able to be more responsive to physical hazards in an environment and make necessary adjustments to avoid things that could damage their gears, motors, cables, etc. and prevent that from happening.
The pain threshold for robots would have to be much higher than the pain threshold for humans. Robots are more resilient and more durable than humans, and their responses would have to reflect that.
However, there is a flaw in the logic that since human reflexes to pain are so effective in preventing injury, a similar process should benefit machines as well. Robots have sensors, humans don’t. Why try to recreate a human system in robots, when robots and humans are so different?
The researchers made the point that humans who aren’t able to sense pain get injured more often than people who can sense pain. Well, a person who was equipped with a series of cameras and sensors would probably get injured even less often than us measly humans with central nervous system. You could argue that it’s better to avoid collision at all than to react to a collision.
However, there could be an advantage for robots to be able to sense “pain” rather than just avoid collisions altogether. Imagine that a fly makes it’s way into a factory. A robot that’s designed to avoid contact with anything that it doesn’t recognize would spend the afternoon trying to evade a harmless fly. A robot that could categorize a physical presence by the likelihood that it could damage the robot wouldn’t be sidelined by a harmless insect.
Maybe robots don’t need to feel “pain”, but the goal of this research is to keep robots safe, and keep humans safe as well. There’s a growing trend in collaborative robots, and the more humans and machine interact, the greater the need for safety measures.