Global urban localization of outdoor mobile robots using satellite images

Abstract

Localization is one of the major research fields in mobile robotics. With the utilization of satellite images and Monte Carlo Localization technique, the global localization of an outdoor mobile robot is studied in this paper. The proposed method employs satellite images downloaded from the Internet to localize the robot iteratively. To accomplish this, the proposed method matches the local laser scanner data with the segmented satellite images. Initial test results conducted on the METU campus are found to be quite promising. Further improvement of this approach has the potential of cutting down not only the operational costs but also the preparation period of the mobile robot enabling researchers to operate their robots in diverse outdoor settings.

Monte Carlo Localization and registration to prior data for outdoor navigation

“…There has also been recent work to localize directly to satellite data. In [8] overhead imagery is segmented into categories corresponding to freespace or a tall obstacle, and then uses a range based observation model similar to those used for indoor localization. A similar approach is applied in [9] but with contours in the imagery as a proxy for obstacle boundaries.…”

Outdoor mapping and localization using satellite images

“…The color codes of four categories are arranged such that the virtual laser scanner operating on the map extracted from satellite images is sensitive to the first two categories, building and dense tree groups and is insensitive to last two categories in the map, that is, field and road. 18 Once P [x t+1 |x t , a t+1 ] and P [y t |x t ] are obtained from the process and measurement models, MCL can be implemented to run the Bayes filter.…”

“…17 Then, the Monte Carlo Localization (MCL) technique is used along with these maps to find the pose of a mobile robot which senses around by a laser range finder. 18 In order to successfully apply the proposed method, satellite view of the objects in scene should correlate with the observation of the mobile agent on ground. In other words, it is expected that the boundaries of the objects present in the map extracted from the satellite images match the observation of the robot in the field.…”

Building lane-graphs for autonomous parking

“…Aerial imagery may become outdated so in-vehicle sensors would need real-time assurance that regions are drivable. In combination with other onboard sensors such as vision sensors and range finders, aerial images have been used for generating cost maps for long-range traversal [16], global localization [7], building and maintenance of robots' world model [13], [17], and mapping [11].…”