Pulse-type hardware neural network mimicking spinal cord function

There is a lot of research on the application of artificial neural networks as a control method for robots to achieve bio-mimetic movements. In this paper, we have designed, fabricated, and measured a pulse-type hardware neural network that mimics spinal cord function. The model was able to change the pulse sequence, pulse width, and pulse frequency according to the input from a sensing model fabricated to mimic the pressure sensation of an organism. As a result, we have succeeded in developing an IC that can perform the same processing as the biological spinal cord. This makes it possible for a quadruped robot to not only change its gait but also to avoid danger by autonomously stopping its walking motion temporarily. The results show that this research has the potential to contribute to the realization of small robots with sophisticated and delicate movements, like living organisms.