Fig. 1:    The relationship between the real world, map, and user’s position and orientation.

Adaptation

Translating between reference frames gives the system the ability to adapt the map display through four different subtasks: panning, rotating, zooming and selection. This adaptation allows the users to quickly learn where they are, thereby freeing them to interact and receive content about specific landmarks of interest.

Adaptive Panning

The process to adapt to changes in the user’s position is flexible in order to avoid unnecessary panning invoked by small unintentional movements. A tolerance area determines what reference frame (i.e., user reference frame or map reference frame) to align to the absolute reference frame. If the user’s position changes within the tolerance value, the user reference frame translates to align itself with the absolute reference frame, moving the symbolic representation of the user’s position on the map reference frame. If the user moves out of the tolerance area, the map refer-ence frame is translated instead.

The tolerance area is a rectangle centered on the screen. Its dimension is defined as a percentage of the width and height of the screen, that is, of the map displayed (Figure 2).

Fig. 2:    Position tolerance of the adaptive panning.

Adaptive Rotation

Sensing the user’s position and orientation allows the map to rotate in accordance with the user’s frame of reference. This adaptive rotation eliminates the north-orientation convention. Two dis-tinct modes of operation are applied: user-centered and map-centered orientation mode. In the user-centered orienting mode, the symbolic representation of the user’s orientation consistently points towards a relative reference, that is, towards the top of the screen. As a result, the map reference frame rotates around the user’s position, thus always aligning itself with the absolute reference frame. In the map-centered orienting mode, the user reference frame is rotated instead.

A threshold value ensures smooth rotation. This threshold is defined as a degree value in the sys-tem’s preferences and depends on the accuracy of the orientation sensor. The threshold defines an imaginary cone in which the orientation may change without causing either the symbol for the user's heading or the map to rotate (Figure 3). Whereas this adapts the map display, the system always knows the exact reading of the orientation. This reading is important for an accurate automated selection.

Fig. 3:    Orientation tolerance of the adaptive orientation.

Adaptive Zooming

Adaptive Zooming is performed based on the context of the user, such as entities of interest and motion speed. The zoom level is different whether a user is walking, riding a bike, or driving a car. The zoom level also changes based on which thematic classification the user is interested in, working with entities such as trees versus lakes or counties. The adaptive zooming functionality is also part of the automated selection method. For example, if the user selects the thematic clas-sification of trees and then the system selects the tree in front of the user the system can zoom in on that objects showcasing information about the object including its relation to the user

Automated Pointing Based Selection

Pointing based selection allows a user to find information about their surroundings just by stand-ing and pointing at the feature of interest. An automated selection queries data with the question “what is that?” In a first step, all the features that intersect the sensed direction from the user’s position are returned. The second step filters this set of features for the feature closest to the user (Figure 4). An extended question of this query is “what Y is that?” In this case, only features of type Y are returned in the first step of the automated selection process.

A challenge of the automated selection is to create a selection that is itself adaptive. This adapta-tion includes a reasoning to adjust the distance from the user’s position in which features are selected. An additional adaptation of the query is to reason about what feature types to select. In the question of “what is that?” an adaptive automated selection has means to extend this query to the question “what Y is that?” based on the user’s context (e.g., mode of transportation, interests).

Fig. 4:    A query for “What is that?” returns X; a query for “What Y is that?” will return Y1.

Use-Case for Implicit Opt-in Advertising

The adaptive nature of the above map interaction is one of three essential parts of a mobile navi-gation aid and tour guide that contains an implicated opt-in advertising model. The real time adaptive panning, rotating, and zooming of the map allows the users to look at their mobile de-vice and quickly translate between what they see in the real world and what they see on the de-vice. By helping to improve the users translate between their perception of their surroundings and the information on the mobile device the cognitive load for the user is decreased so they can spend less time wondering where they are and more time learning about their surroundings.

The second essential part of this system is the content. The ability to point at a building and automatically get content about it enables the distribution content. This is not enough to motivate users to pay to use the device. Neither is an application where a user points at a building to just get coupons. Most users do not get the newspaper for the coupons they get it for the news. But what they pay for the newspaper is not enough for news agencies to survive. The news agencies earn their money from the advertisements. This concept might seem obvious to most people but we need to make sure this concept is carried over to the mobile advertisement domain. To moti-vate users to pay to use a device where they can point at buildings will be to off them rich cul-tural and historic information.

The third essential part of an implicated opt-in advertising system is to have a full-time wireless Internet connection. In a distributed system some of the static content can be cached. For exam-ple, really large audio and video files describing the history of a building, can be stored locally on the device. Other content needs to be dynamic and pulled off the Internet in real-time, such as advertisements, coupons, reviews, and event schedules. This full-time connection also allows for a two-way flow of information. Every time a user points at something their position, pointing angle, and what content they interacted with is stored in the database. This is useful for future analysis.

Imagine a businessman at a conference in a city he has never been to before.  He either rents a device from a visitor center or maybe from his hotel or he could even rent a device beforehand off the Internet. In fact, maybe the cell phone he owns already has this application installed on it. He uses the adaptive map and route guidance functionality to help navigate the city. He sees an old interesting looking building, points at it, and gets an audio description of the historic archi-tecture. He also learns that there is a live band playing in the restaurant on the second floor and if he goes in within the next half hour he will get a free martini with the purchase of any entrée. He also can view the restaurant’s menu on the device.