Navegação inercial veicular de baixo custo : caracterização de sensores, implementação em tempo real, influência do modelo gravitacional e da taxa de atualização

Abstract

With the recent evolution in the field of autonomous vehicles, there is also an increasing demand for more accurate and precise navigation systems capable of composing high-efficiency, safe, and low-cost control systems. Standing out as one of the main navigation solutions employed in such vehicular applications, Inertial Navigation Systems (INS) utilize an Inertial Measurement Unit (IMU). This study presents an inertial navigation algorithm with mechanization in the NED frame as the primary strategy for developing the analyses, as well as some fundamental INS concepts, such as the sensors used, error modeling, initialization, and alignment. Despite its wide applicability, the INS shows a high rate of error growth due to various errors involving the navigation algorithm, sensor error characteristics, and the rate of inertial data transmission for navigation. Therefore, this project presents three studies related to this navigation method: the first performs an analysis of the stochastic nature of inertial sensors using Allan variance techniques, enabling the validation and selection of IMUs based on their stochastic characteristics; the second study analyzes the effect of the gravitational model integrated into the navigation, proposing a gravitational model that provides better accuracy in estimating local gravity acceleration; and the third study involves a performance comparison of an INS regarding the treatment and rate of inertial data transmission provided by the M8U device from the manufacturer Ublox.