Project Reality View
In cooperation with the Geoinformation Group of Potsdam University an Navit extension based on augmented reality will be developed. The Android platform is the basis on which Navit is also available now. Complete instructions, Screenshots and of course the source code will be available in the next weeks at this site. Until then, the following concept introduces the idea of the project.
Dipl.-Ing. Mario Kluge
University of Potsdam
Department of Geography
phone:0049 331 977 2629
Introduction Augmented Reality (AR) combines detailed real-world image data with virtual route information. Clear and easy to follow instructions provide maximum benefit for pedestrians in particular. The pedestrian navigation system called Reality View is based on Navit and uses the existing routing algorithms and menu structure. Displaying superimposed camera images and virtual models with a high degree of congruence requires a number of sensors. Together with GPS an electronic compass and an acceleration sensor are used. Thus equipped, the device is able to recognise the location, position and alignment and adapt the display accordingly. Since establishing itself as a vehicle navigation system, pedestrians are now the focus of current Navit development. As opposed to vehicles, pedestrians posses a much more detailed movement structure. Not just thoroughfares but also paths and areas such as pedestrian zones, parks and open public spaces have to be incorporated into the route guidance. The stock and structure of underlying geodata must therefore be partly regenerated or adapted to the requirements of the target group. The concept is based on the experience and approach of navigation principles in 3D-game simulation. At the screen users have a restricted visual range that is augmented with the aid of additional elements. The basis for comparing 3D computer simulation and the navigation of pedestrians is represented by the augmented perception of the user. Analogous to operation of the camera in mobile telephones, every movement is simultaneously shown on the display. Thus it is possible to have a direct and frontal view of the relevant determination points such as junctions, underpasses or building entrances. The virtual model information required for navigation is segmented and superimposed onto the camera image. With the aid of Reality View (cf. chapter 3) the user’s restricted real visual field is enhanced with prescient route information and presented as Augmented Reality (AR).
Pedestrians Pedestrian and vehicle movement and route choice behaviour differ. Different factors influence route choice. Unlike vehicles, the fastest or shortest route does not always have highest priority. Crucial factors governing route choice include the attractiveness of surroundings, shops and restaurants en route, as well as pedestrian safety.
Navigation methods Since the inception of 3D game simulation, various methods have been developed to enhance user-computer interaction. Irrespective of game genre and tasks set to players, 3D game simulation provides a number of methods to support navigation. Thanks to years of experience in 3D game technology, there are many effective methods available today that assist users of computer navigation. Newly in use is a dynamic cable adapted to the route. Like a wire stretched to the destination, this cable follows the selected route based on a real-world situation.
Reality View An integrated optical camera allows a live stream of the selected perspective to be displayed directly on the screen. The combination of detailed image data and precise route information enables the user to recognise his current environment and at the same time follow the route description. Reality View provides the basis for the the pedestrian navigation system presented here. Implementation requires a mobile device with a graphic display and other components. Absolute positioning in space requires the use of a satellite receiver. Since superimposing the reality view and the route information requires 3 dimensions, a compass and position sensors are also used. Shifting or rotating the standing position can thus be registered by the device and be displayed in the camera scene. Live images of the surroundings required for the reality view are generated using an integrated camera.
Presentation The combined view of camera image and virtual model has to be adapted for superimposition. The principal disparity is in the superimposition of a 2D map perspective and 3D camera image. Even though the camera does not truly generate a 3 dimensional image, it nonetheless shows raised features and therefore view obstructions. A realistic combination of both views thus necessitates modifying the base map to the real-world situation. This is made possible by applying more sensors which apart from calculating the position and orientation of the device also determine its alignment, i.e. location with respect to the horizontal plane. Using this information the navigation software is able to adapt the route information displayed to the actual real-world view. Here the focus is on determining the actual visual range based on the real-world situation. Obstructions on the base map are blended in the maximum visual range of the camera. The result is a reduced visual range adapted to the real-world surroundings.
Base Maps Navigation should be based as far as possible on the use of open geodata, i.e. geodata available for free distribution. The OpenStreetMap project takes centre stage here. The extent to which these data are suitable for pedestrian navigation purposes is looked into. To offset the drawbacks of these jointly developed data, geodata of different origins are consolidated using special data merging algorithms. Data from commercial suppliers as well as that provided by Land Registry offices are integrated. The result is illustrated by a combined dataset enhanced for pedestrian navigation with the aid of separate attributes from different sources. A drawback of jointly collected data is the absence of attribute standardisation. The application of a standard object type index as with data maintained by the Land Registry is a prerequisite for use in navigation. It is therefore necessary to compile such an index for pedestrian navigation purposes. It should specify the source of the assumed attributes as well as the necessary and not yet recorded objects.
Summary In the schematic display, the result shows that the generation of a pedestrian navigation system with Reality View uses a combination of different components. The current prototype comprises a UMPC (Ultra Mobile PC) and the different external components. Since neither size nor weight are ideal, the prototype exclusively serves demonstration and experimental purposes. Upgrading the software to the Android platform is pending following completion of the Linux-based prototype on the Wibrain UMPC. The advantage is in the current varied hardware supplying all the necessary components such as GPS, electronic compass and acceleration sensor. Together with the open and therefore easily enhanceable Android Basis, the ideal environment to provide Navit for pedestrian navigation purposes.