Development of Neuromorphic Locomotion Generator with Switching Gait by Light Stimulation for Microrobots

Kosei Sekiyama, Isuke Okuma, Wataru Nakayama, Shuxin Lyu, Katsuyuki Morishita, Ken Saito

Abstract: The authors are studying the use of analog integrated circuits to mimic biological neural networks and their application to robot control. In our previous studies, an insect-type microrobot equipped with a neuromorphic locomotion generator with an integrated circuit successfully walked with a single gait pattern. Also, we developed a neuromorphic locomotion generator that can switch the robot’s gait. However, the robot had no sensors to receive information from the external environment and change its gait patterns. To implement sensors in microrobots, we focused on receptor cells in the sensory receptors of living organisms. Therefore, the authors developed a receptor cell model that mimics the function of receptor cells to convert external stimuli into electrical signals. In this paper, the authors added a receptor cell model to the same layout of the neuromorphic locomotion generator to support sensory input. Also, we connect an integrating circuit and a photovoltaic cell to the outside of the chip. As a result, we confirmed that the neuromorphic locomotion generator outputs different gait patterns depending on the amount of light stimuli to the photovoltaic cell.