Descrição
The research in this doctoral thesis presents the development and implementation of an estimation scheme, based on a Kalman filter, for the state estimation of vehicles during navigation, fusing measurements from low-cost sensors. For navigation purposes, the states of interest are the components that define the position and attitude of the vehicle. The inclusion of information regarding the dynamics of the vehicle, together with measurements from one or several sensors, enhances the accuracy and precision to which the states can be determined. The proposed state estimation scheme is implemented in pedestrian and aerial applications. For pedestrian applications a methodology is developed, to improve the position estimation by using aerial images. Also, the flight dynamics of precision aerial delivery systems is studied thoroughly, specifically for micro-lightweight parafoil-payload systems. For these systems, two linear models that capture their flight dynamics are proposed and implemented in the state estimation scheme for position and attitude determination, fusing measurements from low-cost sensors employing a Kalman filter. The implementation results proved that the state estimation scheme is effective and suitable for low-cost applications. The incorporation of aerial images in the position estimation process for pedestrian applications allows for the accuracy determination of low-cost navigation sensors, as well as improving the estimation of the position. The lineal models developed for micro-lightweight precision aerial delivery systems proved to capture their flight dynamics. The use of these models, together with the fusion of measurements from low-cost sensors in the estimation scheme, demonstrates an improvement in the position and attitude determination of the vehicles during their descent trajectory.
