Integration of GPS and Accelerometer Uncertainties to Improve the Estimation of the Pose of Autonomous Vehicles
Uncertainty fusion techniques based on Kalman filtering are commonly used to provide a better estimation of the state of a system. A comparison between three different methods to combine the sensor information in order to improve the estimation of the pose of an autonomous vehicle is presented. Two sensors and their uncertainty models are used to measure the observables states of a process: a Global Positioning System (GPS) and an accelerometer. Given that GPS has low sampling rate and the uncertainty of the position, calculated by double integration from the accelerometer signal, increases with time, first a resetting of the estimator based on accelerometer by the GPS measurement is done. Next, a second method makes the fusion of both sensor uncertainties to calculate the estimation. Finally, a double estimation is done, one for each sensor, and a estimated state is calculated joining the individual estimations. These methods are explained by a case study of a guided bomb.