The Biomimetic Real-time Cortex (BioRC) project aims to design, simulate, and construct nano-circuits that model portions of the cortex, where logic and reasoning occur. The nano-circuits designed under this research could find application in autonomous vehicle control, facial recognition, and advance our understanding of biological neural processes. This project also includes development of predictive mathematical models to determine the feasibility of constructing large-scale cortical emulations in hardware. With the brain's large number of interconnections and the need for long term learning, several problems arise in large, real-time, hardware-based emulations; scale, interconnectivity and plasticity. To meet the dynamic needs of such emulations and to deal with interconnectivity and scale problems, the technological requirements of autonomous assembly and reconfiguration are also being evaluated.
Using a bottom up methodology, nano-electronic circuits, including learning apparatuses, from a single neuron to a column of neurons, are being simulated using the latest carbon nanotube and nanowire device models. The circuits implement common neural phenomena and mechanisms, including ion channels, dendritic computation, spike bursts and spike trains. Circuits are being designed with the ultimate goal of scaling to cortical-sized networks, to be implemented in future 3-dimensional nano-circuits and with awareness of possible dynamic hardware technologies including autonomous assembly and reconfiguration. In the near term, simplified neural circuits are being constructed using nanotubes and nanowires. Bottom-up chemical synthesis and top-down nanofabrication are combined to assemble nanowires and carbon nanotubes into hierarchical structures to be used as building blocks for cortical circuits. Ancillary to this activity is the development of mathematical models of neural size and interconnectivity that are used to evaluate solutions to the interconnection problem and determine feasibility of a synthetic cortex.
