Search

In this study, Bayesian Filter based reliable localization techniques that use depth information collected by range measuring sensors in indoor/outdoor environments, or terrains in general, are commonly analyzed and sample applications to both land and air vehicles have been demonstrated in real and simulated environments. For the applications, a reference depth map of the mission space is generated first, if not available already. This map is used by the vehicle for localization purposes. When the mission starts, the vehicle does not know its whereabouts and it is assumed that GPS-like external localization systems cannot be used. The vehicle compares the observations coming from onboard range sensors like Sonar, Lidar and/or Synthetic Aperture Radar (SAR) against the reference map and updates its estimated position at every step. The estimated position and the actual position equates after enough observations are made. In the first sample application presented, a Pioneer P3- DX model research robot determines its indoor position using one-dimensional localization. In the second application, an Unmanned Aerial Vehicle (UAV) performs localization in a simulation environment using synthetic two-dimensional elevation data. And in the last example, UAV localization is demonstrated again using real Digital Terrain Elevation Data (DTED). A discussion about faster and more efficient localization in case of the presence of additional localization data is also presented. [ABSTRACT FROM AUTHOR]