This article explores the use of equivariant filtering in inertial navigation systems to improve state estimation in uncrewed aerial vehicles (UAVs). Traditional estimation methods, such as the multiplicative Kalman filter (MEKF), have limitations in their consistency, initial state estimate errors, and convergence performance. The authors highlight the advantages of equivariant filters (EqFs), which exploit the system’s mathematical properties and offer faster convergence rates and robustness against wrong initial state estimates. The article demonstrates the effectiveness of EqFs using GNSS sensors and an inertial measurement unit (IMU) in real-world scenarios, showcasing its potential for enhancing UAV state estimation.
Publication date: 18 Oct 2023
Project Page: DOI follows ASAP IEEE
Paper: https://arxiv.org/pdf/2310.10597
