Soft Error Assessment of Attitude Estimation Algorithms Running on Resource-Constrained Devices Under Neutron Radiation

There is a growing incorporation of unmanned aerial vehicles (UAVs) within remote and urban environments due to their versatility and ability to access hard-to-reach and/or congested places. UAVs offer low-cost solutions for many applications, including healthcare (e.g., medical supplies delivery) and surveillance during public events, protests, or emergencies (e.g., a nuclear accident). However, drone utilization in urban areas often relies on strict regulations to ensure safe and responsible operation. UAVs are subject to radiation-induced soft errors, and identifying the most vulnerable software and hardware components to radiation exposure is an advisable task, which is difficult to undertake. An essential task to UAVs correct operation is attitude estimation (AE). This article assesses the soft error reliability of three AE algorithms running on two resource-constrained microprocessors under neutron radiation. Results suggest that the extended Kalman filter (EKF) algorithm provides the best Mean Work to Failure (MWTF) result for critical fault events, which is about
$3\times $
more than the indirect Kalman filter (IKF) and
$1.5\times $
more with respect to the novel quaternion Kalman filter (NQKF) algorithm.