Estudo da reconstrução de trajetórias baseado em sensores inerciais de baixo custo no contexto de mobilidade terrestre

Abstract

The localization of a person allows to perform several tasks with regard to the concept of mobility. It enables a greater range of possibilities related to infrastructure and services in a city, in particular concerning the Intelligent Transportation Systems. This work proposes a methodology for reconstruction and identification of trajectories, using low cost inertial sensors. The proposal includes the preparation of the acceleration signal for a double integration stage, in order to obtain the displacement in the measuring axes. For the signal conditioning, we have used a FIR type digital filters and the mean values along a time window. In view of avoid errors in performing the double integration, due to certain "contaminations" (noises and offset) that tend to remain in the signals even after the use of filters, a function called pause detector was proposed. This function is performed together with the whole system, aiming to zero the speed in the integration, and consequently to pause the displacement as soon as a pause is detected. An analysis of the quality of the data returned by the sensors and the results obtained is carried out, as well as an observation of the consequences of respecting or not the definition of Nyquist's Theorem. After that, we have proposed a solution for improve the results by using a cubic splines.