An artificial neuromuscular junction for object tracking

Real-time object tracking is required for demonstrating artificial vision systems. However, conventional silicon-based CMOS neuron systems are improper because they cannot provide sufficient output current to modulate the motor, which is for adjusting the position of the physical artificial eyeball. Hence, we propose an artificial neuromuscular junction (NMJ) employing a GaN HEMT to control the position of the artificial eyeball. Due to the high carrier mobility of GaNHEMT, the NMJ drives an adequate output current for modulating an artificial eyeball. A ferroelectric film is deposited on the GaN HEMT as a gate dielectric to emulate the bio-inspired behavior, specifically long-term plasticity for the learning process. Furthermore, this design aims to develop a GaN ferroelectric HEMT (FeHEMT) based array circuit system for processing the data from vision system and immediately react at impetus through the in-situ neural network.