[Back]

L. De Cock:
"Adaptive Mobile Indoor Route Guidance, The Next Big Step";
Supervisor, Reviewer: A. De Wulf, P. Bogaert, L. De Marez, E. De Poorter, G. Gartner, H. Huang, L. Meng; Department of Geography, Universiteit Gent, 2021; oral examination: 2021-10-08.

Outdoors, we have a lot of options for navigation. We can use several apps on our smartphones, calculate the route for a vehicle or another transport mode, choose the time and date for which the route has to be calculated, avoid highways, estimate traffic and get notifications of speed controls. We
can even change the voice in our app so that Batman can tell us we have reached our destination. Have we though? When we for example have arrived at the hospital entrance, have we really reached our destination? We still have to find the reception desk, we have to get to the waiting room and maybe grab a coffee on the way, but once we enter the building we´re on our own. No apps that can say how long it will take, no funny voices to make
us feel more comfortable.
In a world where many new construction projects have a digital twin, where not only people but also devices are connected through the internet of things, where sensors and cameras are generating big data and where everything has to be smart, indoor navigation is the next big step. However, this doctoral dissertation does not focus on the technological implementation of indoor navigation. Instead, it focuses on the cognitive aspects of navigation in smart buildings, and more specifically on easing the
decision making process during route guidance. As every decision point is different, the user´s need for route information is also different at every point. Therefore, decision making can be eased by adapting the route instruction type to the needs of the users, and thus, to the decision point.
Navigation systems that implement this idea provide the right amount of information at the right time and place. The usability of these systems is studied in this dissertation.
The first step in the design of an adaptive navigation system is to determine which route instruction types should be used on which decision points. In this research, this decision is based on the subjective preferences of the users, which was collected during an online survey. The case study building
of this work is the iGent, the office lab of Ghent University, and for the online survey ten route videos were recorded in this smart building, and ten route instruction types were designed for these route videos (e.g., maps, symbols, photos, 3D-simulations). Participants had to indicate how complex
they found a decision point and how they scored a route instruction type on every decision point of the recorded routes. The results indicated, first of all, which decision point categories were found to be most complex, and how this could be related to the building configuration, quantified by space syntax. Second of all, they indicated which route instruction type gained
preference on which decision point category.
In a second step, the results of the online survey were used to develop a mobile indoor navigation prototype. The prototype was web-based, connected to the UWB sensors in iGent and automatically showed a new route instruction on the smartphone of the users. This route instruction could either be adapted to the decision point (i.e. symbols at starts and ends, 3D-simulations at complex turns and photos at all other points) or not
adapted (i.e. photos on all points). The usability of the adaptive and non-adaptive system was tested with objective measures (e.g. eye tracking) in a field experiment where participants had to walk three routes with either one
of the systems. The results of this field experiment showed that the usability of the adaptive system was higher, both in terms of cognitive load and performance.
In a third step, a virtual model of a building floor was designed and a virtual copy was made of the adaptive prototype from the field experiment. The same parameters as in the previous step were used to test the usability of the prototype, but this time the experiment was conducted in virtual reality.
Because the virtual model was much bigger than the physical building, an analysis on the building configuration could be included. This way, the results of both the online survey and field experiment could be cross-validated in virtual reality. The results showed that the complexity perception of the turns matched the rise in cognitive load of the participants, but only with a photo instruction this influenced the speed of the participants. At end points, only the cognitive load induced by a symbol instruction could match the complexity perception of participants. On the
whole, the lower cognitive load of the adaptive instructions found in the field experiment was confirmed in the virtual reality experiment.
The results of this research were translated into practical guidelines or implications for the design of adaptive mobile indoor route guidance systems, because this work has shown this is the way to go.