Roll angle estimation of airborne vehicles using the minimal array of atmospheric temperature sensors

The use of low-cost thermopile sensors for aerial vehicle roll angle estimation was investigated as an alternative to GPS reliant navigation solutions. The minimal number of measurements necessary to estimate vehicle roll was determined. A physics-based sensor model was used to investigate the effect on estimate error over sensor field-of-view angle and for nonminimal sensor sets. The bound on error introduced by the reduced-fidelity sensor model used for estimation was detailed and discussed. The performance of the minimal, three sensor array was contrasted with a larger array and demonstrated that sub-degree accuracy is possible with the method of estimation introduced here.