Data Collection and Analysis
Wizard of Oz study of indoor positioning technologies at Paris training house, March 2007
In the second mission, we provided the commander of the mission with a floor plan marked with the position of the bottles to be found. The commander and the reconnaissance team communicated with each other with 2-way radio devices. Using this channel, the commander guided the team inside of the building. The team, on the other side, provided information about their environment that the commander translated to the floor plan and used to find the current position of the team. By means of this change, some very interesting effects in the praxis emerged to accommodate and take advantage of the new information. The reconnaissance mission ran no longer systematic, as they followed the directions of the commander and went direct to the places were the bottles were located. The radio communication was of very low quality, and for these reason, the team inside the building used some very basic expressions to identify and describe their position to the commander. By the identification of the disposition of the doors and walls, it was possible for the fire-fighters to collaborate with the commander in finding out the position of the fire-fighters. Even though this location was not perfectly clear for the commander, through the use of landmarks for navigation, he was able to provide a quite direct guidance.
In the third mission, the commander used the command post application to follow the fire fighters on a large interactive display using the position information provided by the Wizard of Oz platform (see figure below). As in the previous mission, the commander and the team were communicated through 2-way radio. This time there were no doubts about the actual position of the fire-fighters, so the guidance was by means of directions. The Wizard of Oz platform, however, didn’t provide heading information, so it was not possible for the commander to have a reference indicating the direction of the next movement. This problem was also solved after some trial and error by an emerging collaborative praxis. He started to ask the fire-fighters to move forward and then backwards to identify the direction of movement. Based on this information, he was then able to give instruction like “go forward 4 meters” or “find the next door and then left”.
The change on the navigation mode of the fire-fighters from finding their own way in the first mission to being guided in the second and third mission had a direct impact in the report of the fire fighters. When guided by the commander, their ability to recall the path and the de-tails of the building were considerable lower. This can be understand as a negative effect of the use of location information for guiding the fire-fighters, but it is also very difficult to tell if such effects can be overcame by expert training around the new available practices. These aspects are questions for our future research activities.
Evaluation of Command Post Application, December 2006/ March 2007
Results of the test of the Command Post Application are shown in the figure below. Detailed results of the December study are available in the Diploma Thesis of Stefan Habelski, “A Design Study in 3D User Interface for Large Interactive Displays“, defended at the University of Magdeburg, Germany.
Results of the PDR evaluation are shown in the Figure above. The figure shows positioning accuracy of both PDR and GPS positioning data. Due to magnetic perturbations inside buildings (from the building structure and from electrical equipment), obtaining stable heading information is very difficult. This is the subject of current research.
Sensor Network LifeNet
The results of the measurements of the Relate sensor nodes are reported in the Relate Deliverable 7a. As these studies are preparatory technical studies in order to validate and tune the positioning accuracy of the sensor network without an immediate impact on the users, they are not repeated here. The essential result from the studies is that the positioning accuracy is sufficient to provide useful navigation support to the fire fighters thus validating further development in this direction.
Virtual Reality Simulation FireSim
The study identified a number of shortcomings in the usability and more general user acceptance of the application that have to be addressed before conducting a formal evaluation of the LifeNet. These include improved steering behaviour for the character, better feedback for navigating along walls and opening doors. Interestingly, the fire fighters consistently asked for the possibility to kneel down during the reconnaissance missions which is part of their standard work practice. What is interesting about this is that this shortcoming did not prevent the players from using the application but provoked quite determined objections, indicating that this and similar features will be important for user acceptance of the application.
Besides these usability issues, the application was well received and created quite a lot of attention, attracting a considerable number of fire fighters that we did not have to invite or ask to take part in a simulation. After playing a few sessions it also became apparent that variation in the provided scenarios is important to maintain the interest. Several suggestions were made to make the application more appealing like for example introducing a high score and hall of fame.
In terms of the LifeNet, the players readily understood the concept and started using it for their navigation. During this usage several problems became apparent that are partly related to the LifeNet implementation and partly to its virtual simulation. For example, we noticed that at times there was interaction between nodes through the ceiling between different floors. This is a simulation artefact that we will eliminate for the next study. But specifically at doors, the navigational information provided to the fire fighters seemed to be ambiguous and the players did not always know how to interpret them. This has sparked a further reflection about the way the navigation information has to be presented and has let to new versions of the interface.
