Sensors are usually not responsible for processing information, only sending data to a single, large, powerful central processing unit (CPU) where learning behavior occurs. Therefore, the existing robots are usually cumbersome in wiring and delay in response time. They also have a glass heart that is very vulnerable to injury..
One of the authors, arindam Basu, said: one of the concerns is how to make sure that robots can interact with us safely when we work with robots. Therefore, scientists all over the world have been looking for ways to make robots have the ability of perception, such as feeling pain, reacting and withstanding harsh working conditions. However, the complexity caused by the combination of many sensors and the vulnerability of complex systems themselves are the main obstacles to the widespread adoption of traditional methods.
First, the team innovated memtransistors, which function as multiple synapses in human brain neurons, enabling neural networks to have thousands of similar connections.
They then built the memristor transistor into satellite threshold adjusting receivers (stars), making it a brain like electronic device capable of remembering and processing information. In this system, they are used as AI pain receptors and synapses of robots.
The self-healing features of these new devices can help robotic systems stitch wounds repeatedly, even at room temperature, in the event of a scratch or scratch, said Rohit Abraham John, one of the authors. Its similar to how our biological systems work, just as human skin heals itself.
In a public video, the team showed how robots learn to react to injuries in real time.
They first unplugged the robots wires.
Finally, as for its contribution to the progress of science, the robot in charge said: ud83dude42ud83dude42ud83dude42ud83dude42ud83dude42ud83dude42ud83dude42ud83dude42ud83dude42ud83dude42ud83dude42ud83dude42ud83dude42ud83dude42ud83dude42ud83dude42ud83dude42
Source: Academic headlines editor in charge: Mao Xinsi_ NBJS11624