Personal Wayfinding Assistance

We are traveling many different routes every day. In familiar environments it is easy for us to find our ways. We know our way from bedroom to kitchen, from home to work, from parking place to office, and back home at the end of the working day. We have learned these routes in the past and are now able to find our destination without having to think about it. As soon as we want to find a place beyond the demarcations of our mental map, we need help. In some cases we ask our friends to explain us the way, in other cases we use a map to find out about the place. Mobile phones are increasingly equipped with wayfinding assistance. These devices are usually at hand because they are handy and small, which enables us to get wayfinding assistance everywhere where we need it. While the small size of mobile phones makes them handy, it is a disadvantage for displaying maps. Geographic information requires space to be visualized in order to be understandable. Typically, not all information displayed in maps is necessary. An example are walking ways in parks for car drivers, they are they are usually no relevant route options. By not displaying irrelevant information, it is possible to compress the map without losing important information. To reduce information purposefully, we need information about the user, the task at hand, and the environment it is embedded in. In this cumulative dissertation, I describe an approach that utilizes the prior knowledge of the user to adapt maps to the to the limited display options of mobile devices with small displays. I focus on central questions that occur during wayfinding and relate them to the knowledge of the user. This enables the generation of personal and context-specific wayfinding assistance in the form of maps which are optimized for small displays. To achieve personalized assistance, I present algorithmic methods to derive spatial user profiles from trajectory data. The individual profiles contain information about the places users regularly visit, as well as the traveled routes between them. By means of these profiles it is possible to generate personalized maps for partially familiar environments. Only the unfamiliar parts of the environment are presented in detail, the familiar parts are highly simplified. This bears great potential to minimize the maps, while at the same time preserving the understandability by including personally meaningful places as references. To ensure the understandability of personalized maps, we have to make sure that the names of the places are adapted to users. In this thesis, we study the naming of places and analyze the potential to automatically select and generate place names. However, personalized maps only work for environments the users are partially familiar with. If users need assistance for unfamiliar environments, they require complete information. In this thesis, I further present approaches to support uses in typical situations which can occur during wayfinding. I present solutions to communicate context information and survey knowledge along the route, as well as methods to support self-localization in case orientation is lost.