Polarized skylight navigation.

Vehicle state estimation is an essential prerequisite for navigation. The present approach seeks to use skylight polarization to facilitate state estimation under autonomous unconstrained flight conditions. Atmospheric scattering polarizes incident sunlight such that solar position is mathematically encoded in the resulting skylight polarization pattern. Indeed, several species of insects are able to sense skylight polarization and are believed to navigate polarimetrically. Sun-finding methodologies for polarized skylight navigation (PSN) have been proposed in the literature but typically rely on calibration updates to account for changing atmospheric conditions and/or are limited to 2D operation. To address this technology gap, a gradient-based PSN solution is developed based upon the Rayleigh sky model. The solution is validated in simulation, and effects of measurement error and changing atmospheric conditions are investigated. Finally, an experimental effort is described wherein polarimetric imagery is collected, ground-truth is established through independent imager-attitude measurement, the gradient-based PSN solution is applied, and results are analyzed.