UCLA

The human brain, a remarkably efficient system operating on minimal power, excels at processing complex sensory data and learning from experience. Inspired by this biological marvel, this dissertation reports the development of dedicated neuromorphic computing hardware, specifically focusing on a 20 by 20 synaptic resistor circuit based on aluminum oxide and titanium silicide material. By mimicking the brain's synaptic connections, these devices offer low-power, non-volatile memory and the ability to learn and adapt in real time, demonstrating the potential of the synaptic resistor circuit through a successful application in drone navigation. These findings pave the way for energy-efficient and adaptable neuromorphic computing systems with broad implications for artificial intelligence.