NAVVIS system guides users through buildings via a collection of photos
While we’ve pretty much got to the point where we take GPS-enabled navigation systems for granted, there’s one rather prominent place where they can’t access the necessary satellites – indoors. A number of projects have taken on the challenge of indoor navigation, incorporating things like accelerometers, “invisible” landmarks, rapidly flickering lights, and magnetic fields. The experimental new NAVVIS system, however, utilizes a database of photographs.
The technology is being developed by researchers at the Technical University of Munich.
They started by wheeling a trolley down the corridors of their maze-like main campus. That trolley was equipped with two laser scanners, multiple DSLR cameras, and a 360-degree camera. As the trolley moved, the cameras snapped photos, while the scanners recorded the horizontal and vertical dimensions of the corridor.
The data gathered by the scanners was used to create a continuous three-dimensional map of the corridors, which was overlaid on photos of the corresponding areas. The result was a database of location-tagged 3D still images.
When test subjects subsequently tried to find their way through the campus, they would periodically stop and take photos with their smartphone, which was running a custom app. By comparing each smartphone photo to those in the database, the app was then able to figure out their location down to the nearest meter, along with the direction in which they were facing – all within a fraction of a second. Based on that information, it would then advise them on how to proceed to their destination, via overlaid 3D directional arrows on their smartphone screen.
To keep the database from becoming obsolete as the corridors change, the user-supplied photos are now incorporated into the database, overwriting older, more dated photos of the same areas.
Although the NAVVIS system is still in development, the researchers are already envisioning other uses for it. These could include augmented reality apps, that would recognize locations in buildings such as museums, and provide users with information on the things that they photographed.
Source: Technical University of Munich