Neural-Inspired sparse wing sensors for insect flight

Controlling high-dimensional systems with fast and robust feedback is a central challenge in the design of many modern engineering systems. This work is inspired by flying insects, which use a few embedded strain-sensitive neurons to achieve rapid and robust flight control despite large gust disturbances and a highly unsteady environment. Even more remarkable is how they achieve this with limited brain capacity and latency in communication. We investigate the sensing of exogenously induced body rotations by biological strain sensors on the wings of insects. Combining a structural wing model and electro-physiological wing neuron recordings, we show that arrays of neural-inspired sensors are able to detect exceedingly small differences in wing deformation that result from body rotations. Furthermore, we used sparse optimization to show that very few sensors in optimized locations on the wing are required to accurately classify rotation. The combination of temporal filtering with nonlinear neural thresholding and the optimized spatial distribution of biological sensors is crucial to achieve fast and robust, yet computationally efficient, sensory feedback. We believe this paradigm holds great promise for engineering design of high dimensional control problems.