Integração de Sistemas de Navegação Inercial com Atualizações Estacionárias e Restrições de Movimento

Abstract

Inertial Navigation Systems (INSs) are extremely important for land, air, space, marine, and underwater vehicular applications. They provide a complete navigation solution (position, velocity and orientation) for the vehicle, at high sampling rate and high bandwidth, being also independent of external signals. Nevertheless, INSs have the disadvantage of accumulating errors in the long term, due to the numerical integrations carried out internally. One solution to mitigate this error accumulation is to aid the INS with auxiliary sensors, usually via Kalman filters. An alternative approach is to use vehicle’s Motion Constraints (MCs) and/or Zero Updates (ZUPTs), as “virtual sensors” in order to aid the INS. This paper investigates the aforementioned problem and shows that the proposed aiding approach is able to greatly reduce INS error accumulation over time. As main contribuition one demonstrates that individually MC and ZUPTs may not be effective in solving the problem of divergence of the terrestrial vehicular INS solution, but that, when combined, they allow a non-divergent navigation with metric level acuracy. Results from real experiments confirm the adequacy of the outlined verifications.