SMA actuator and pulse-type hardware neural networks IC for fast walking motion of insect-type MEMS microrobot

In this study, the actuator system and an integrated circuit (IC) neural network for a hexapod locomotive microrobot is discussed with the aim of achieving walking motion. Miniature size robots such as insects are a field of interest in robotic studies. However, the realization of such a small robot, especially with micro-mechanisms and an artificial brain, similar to those of real organisms, is difficult. We propose the use of micro-electro mechanical systems (MEMS) for the micro-mechanisms and pulse-type hardware neural networks for the brain of a microrobot. The silicon structural component was produced using MEMS. Shape memory alloy-based artificial muscle wires were used for the rotational actuator, and gait motion was generated using link mechanisms. We compared two types of rotors with different shapes and thermal capacities. By optimizing the design of the rotary actuator and the driving current, the robot exhibited faster and stable walking motion. Furthermore, we developed an IC for pulse-type hardware neural networks to act as an artificial brain for the gait controller and the bare chip was connected to the MEMS robot.