Programmable Logic Construction Kits for Hyper-Real-Time Neuronal Modeling
Neural Computation
Hi-index | 0.00 |
An often cited advantage of neuromorphic systems is their robust behavior in the face of operating variability, such as sensor noise and non-stationary stimuli statistics inherent to naturally variable environmental processes. One of the most challenging examples of this is extracting information from advected chemical plumes, which are governed by naturally turbulent unsteady flow, one of the very few remaining macroscopic physical phenomena that cannot be described using deterministic Newtonian mechanics. We describe a "synthetic moth" robotic platform that incorporates a real-time neuromorphic model of early olfactory processing in the moth brain (the macro-glomerular complex) for extracting ratiometric information from chemical plumes. Separate analysis has shown that our neuromorphic model achieves rapid and efficient classification of ratiometrically encoded chemical blends by exploiting early phase chemosensor array transients, with execution times well beyond biological timescales. Here, we test our neuromorphic synthetic moth in a naturally turbulent chemical plume and demonstrate robust ratiometric communication of infochemical information.