6. Conclusions and Future Work

Individuals in unknown locations, such as utility workers in the field, soldiers on a mission, or sightseeing tourists, share the need for an answer to two basic questions: “Where am I?” and “What is in front of me?” Because such information is not readily available in foreign locations, aids in the form of paper maps or mobile GISs, which give individuals an all-inclusive view of the environment, are often used. The panoptic view of these maps may impede the user’s positioning and orienteering process, since people perceive their surroundings in a perspective way from their current position. This paper describes a novel framework that resolves the problem of finding the correct reference frame by applying sensors that gather the individual’s spatial frame of reference. This spatial frame of reference, in combination with an egocentric spatial data model, enables an injective mapping between the real world and the data, hence alleviating the individual’s cognitive workload. Furthermore, this egocentric spatial data model allows information systems to capture the notions of here and there and, consequently, provides insight into an individual’s surroundings. Finally, this framework, in conjunction with the context given by the task to be performed, enables information systems to implicitly answer questions with respect to where, what, and how things are occurring in the user’s surroundings.

6.1. Summary

Paper maps, which provide users with an allocentric view of their surroundings, are often difficult to use, because they are static and inflexible. The inflexibility of paper maps creates three key problems: (1) users have to cognitively place themselves into map space, (2) the user cannot change the thematic information displayed on a map, and (3) users cannot change the granularity or zoom level of the map without creating a new map. Digital maps and GISs have alleviated some of these problems. In a GIS the depicted space is more malleable, the user can pan and zoom over the map.  Some of the problems associated with a paper map still exist in the GIS environment. For example, users who want to find information about their local environment still need to cognitively place themselves into the map space. A problem of GISs beyond those associated with paper maps is that GISs are used in desktop computing environments, where interaction with the computing device is the user’s primary task. Users in the field, such as tourists or tax assessors, are mainly interested in finding information about their surroundings. A direct interaction with their surroundings means that interacting with the mapping system becomes a secondary focus. This makes standard GIS manipulation in the field complicated and inefficient, since a user in the field has a different focus than a user at a desk. Finding an innovative solution to this problem lead to research on intelligent mobile GISs, finally resulting in this paper.

Intelligent mobile GISs are aware of the user’s spatial context and can automatically provide map information in a form the user wants. An intelligent mobile GIS senses the user’s movements and decides what to center the map on, how to orient the map, and what zoom level or granularity best supports the user’s task. The system intelligently decides on these map attributes by sensing the egocentric reference frame of the user. This reference frame is based on the user’s location, orientation, speed, path traveled, and queries made. Once the map data is aligned with the user’s reference frame it is also possible to process egocentric queries.

An intelligent mobile GIS is comprised of a collection of spatially-aware mobile client devices and an egocentric spatial data model. This paper examined two kinds of client devices: (1) a mapping device, which looks like a digital map but has a context driven intelligence, and (2) a pointing device, which acts like a computer mouse for the real world. With the pointing device users can point at objects in their surrounding, click a button, and receive information about the selected object. Users are still communicating with a geospatial information system, but they are not aware of this interaction since it occurs on an application server. Instead of working directly with the map, users of the pointing technology receive attribute information about the object they selected as video, audio, or text. These two mobile computing devices make it easier for users to find information about their surroundings. The second part of the intelligent mobile GIS is a data model. The egocentric spatial data model translates between the allocentric data in the database management system and the egocentric point of view users have of their surroundings.

An egocentric spatial data model has two components: (1) ADTs that store up-to-date data about a user’s position and orientation, and (2) procedures to relate the egocentric reference frame created by these ADTs to the absolute reference frame of the spatial dataset. Such procedures provide the information system with an understanding of the notion of here and there. An information system with an insight into the user’s egocentric perspective can use this contextual information to answer questions users might have about their surroundings with a grammar that is easier for the user to understand. Consequently the interaction between user and information system is similar to interaction between humans.

6.2. Findings

In developing the egocentric spatial data model several thing were learned:  

Spatial databases can process the term “here” and “there.”

This processing can occur with the use of egocentric ADTs that function with up-to-date sensed spatial data.

Location sensors enable new types of spatial queries.

New query paradigm automated based on sensors.

Integration of spatial aware sensors with a query language allows for mobile queries that are transparent to the user.

6.3. Future Work

The egocentric ADT developed by this research allows for possible future research tasks. A variety of issues remain to be resolved. One of them is to study the effect of spatial-awareness on the design of dynamic graphical user interfaces. The use of egocentric spatial-awareness in database management system is an emerging field and many questions are open. The next sections discuss new questions that became apparent through the results of this paper. They address ADT implementation, sensed data histories, context-aware human-computer interaction, privacy issues, ontology-driven user profiles, multimedia integration, and digital terrain models.

6.3.1. ADT Implementation

This paper develops the syntax, semantics, and execution model for the egocentric spatial data model the next step is to implement the ADT within a database management system. Once this system is implemented it will be possible to examine query optimization procedures. With an implemented system it will also be possible to develop query evaluation models, and investigate use case studies.

6.3.2. Egocentric Histories

The egocentric spatial data model allows for a users sensed position and orientation to be incorporated into a database management system. It will be possible represent more than the current positional and orientation data. The system can contain past sensed data thereby allowing the system to track the user. An insight into where the user has been will make it easier to predict where they are going. This will allow the system to pre-buffer data that it deems necessary for the user. Keeping track of the sensed data will allow the system to perform root mean square analysis and improve the data’s accuracy. Some of the difficulties of developing a sensed data tracking system would be; to decide how much history to keep track of, how to represent, and index the historical data.

6.3.3. Context-aware Human-Computer Interfaces

The criteria for creating intelligent mobile GISs, are based on human-to-human communication, such as between a traveler and a Cicerone. When people talk to each other, they relate their information transmission to the context of the situation. This context can be spatially, temporally, and culturally based. In order for a computer to be able to provide a similar level of communication it needs to be spatially-aware. This spatial awareness will allow the system to modify its Graphical User Interface (GUI) to better serve the user’s needs. In this paper we look at driving a map’s centricity, orientation, and zoom level based on the user’s spatial context. In the future the information system interface should be driven by context. This context driven interface more than the GUI, the system should be able to use context to decide what, when, and where to use sound, video, and text to present its information to the user. The interface will be media rich. The basic principles developed in this paper can be used as a platform for research in human-computer interaction.

6.3.4. Ontology-Driven User Profile

Chapter 2 argues that mobile maps, as well as computers, need to become ubiquitous in order for the map to disappear from the user’s conscious perception. One way to make a technology ubiquitous is to build the technology based on an insight into the context of the tasks a user is performing. This contextual knowledge allows the technology to adapt to the needs of the user. The focus of this paper was to collect and use information about the user’s spatial context. In the future, information systems will need to collect other attribute information about the user, such as their likes, dislikes, historical experiences, education, and training. A system with this kind of knowledge will be able to provide information with an increased level of usability to a user. An ontology-driven user profile query system will be able to semantically align users’ informational needs with the information describing their environment. Spatial information is very diverse and collected and stored by different organizations. An ontology profile system will be able to examine the semantics of these diverse data sets and filter the information so that it pertains to the users’ interests. This ontology profile system can intelligently decide on the level-of-detail to provide to the user.

6.3.5. Privacy Issues

There are two key privacy issues that have arisen from the research in intelligent mobile GISs: the first key issue is related to the privacy of the system’s user, and the second one relates to information that can be accessed by the system. The egocentric spatial data model holds information about users that they might want to keep private, for example their geographic position and orientation at any point in time. From this stored information people will be able to know where, when, and what the system’s user is looking at. This information is necessary to accurately aid the user, but it must be secure and private. As using other context information about the user extends the system, the privacy problem becomes even bigger. The second issue relates to the geospatial information that users of the system could have access to. The pointing technology should allow different levels of access, for example, pointing at someone’s house, a police officer or tax assessor should have access to information about the land parcel and house owner, but other people should not. Privacy should be an important consideration when developing the egocentric spatial data model because in order for a mobile GIS to be considered intelligent users must trust the information system. This trust is created though a privacy structure used by the information system.

6.3.6. Multimedia Geo-Footprint Digital Earth

Both the mapping and pointing technologies have the ability to provide non-spatial information. This information can be in many formats, such as audio, video, text, and multimedia. These multimedia datasets need to be linked to the selected features of interest, which could be points, lines, or regions. One problem is that the system might not know what multimedia file to access when a user points at a road or building. Future research needs to examine the development of geo-encapsulation or creating geo-footprints of data. This footprint allows media to link to any geospatial feature. The geo-footprint needs to travel with the media file. Having a separate meta-data file will not work because this information needs to be updated when the file is. Some research has already been conducted in this area of geo-libraries, but it needs to be extended to include all media formats, as well as, being encapsulated in the media themselves.

6.3.7. Digital Terrain Models

An intelligent mobile GIS client uses sensed data about its users’ egocentric spatial reference frame to calculate the features of interest for them. This selection process is based on a terrain intersection model algorithm (Faisal 2003), where a ray from the user’s position intersects a three-dimensional object in a terrain model. These terrain models need to be robust and their file size needs to be small so they can be transferred fast and efficiently. For a terrain model to be useful in such an environment it needs to store information about physical terrain, which is a continuous field data structure as well as data about physical geospatial objects such as buildings and trees. In order for a terrain model to function at many different levels of detail it also need to be database driven. Today’s digital terrain models do not live up to these requirements yet and more research in this area has to be done.