The early design phase for a new building is a crucial stage in the design process of architects. It has to be ensured that the building fits into the future environment. The Collaborative Design Platform targets this issue by integrating modern digital means with well known traditional concepts. Well-used styrofoam blocks are still cut by hand but are now tracked, placed and visualized in 3D by use of a tabletop platform and a TV screen showing an arbitrary view of the scenery. With this demonstration, we get one step further and provide an interactive visualization at the proposed building site, further enhancing collaboration between different audiences. Mobile phones and tablet devices are used to visualize marker-less registered virtual building structures and immediately show changes made to the models in the Collaborative Design laboratory. This way, architects can get a direct impression about how a building will integrate within the environment and residents can get an early impression about future plans.

Gaze tracking facilities have yet mainly been used in general for marketing or the disabled and, more specifically, in Augmented Reality, for interaction with control triggers, such as buttons. We go one step further and use the line of sight of the user to attach information. While any information may not conceal the view of the user, we displace the information by an angular degree and provide means for the user to capture the information by looking at it. With such an apporach we see a potential for faster resuming times of the original task for which a required information needs to be accessed. The demonstration shows a comparably complex primary task assisted by our gaze-mounted information and illustrates the inherent differences for information access w.r.t. conventional methods, such as listing action items at a fix position in space or on a screen.

Integrating different mental models and the corresponding interaction techniques for navigation and object manipulation tasks is a topic of concern for providing user interfaces for the wide variety of potential users. With a user controlled, egocentric hand-held device we aim at integrating the so far identified three major interaction techniques based on the metaphors of a steering wheel, a toy airplane and a window frame. Here, specifically the toy airplane and the window frame contradict each other, leaving the issue at which postural angle of the hand-held device the mental model of the users changes.

The calibration of optical see-through head-mounted displays is an important fundament for correct object alignment in augmented reality. Any calibration process for OSTHMDs requires users to align 2D points in screen space with 3D points in the real world and to confirm each alignment. In this poster, we present the results of our empiric evaluation where we compared four confirmation methods: Keyboard, Hand-held, Voice, and Waiting. The Waiting method, designed to reduce head motion during confirmation, showed a significantly higher accuracy than all other methods. Averaging over a time frame for sampling user input before the time of confirmation improved the accuracy of all methods in addition. We conducted a further expert study proving that the results achieved with a video see-through head-mounted display showed valid for optical see-through head-mounted display calibration, too.

Systems using two-handed spatial manipulation techniques also require strategies to enable system control tasks. These strategies make it possible to interact with the system comfortably while controlling two hand-held objects simultaneously. The Augmented Chemical Reactions project makes intense use of such two-handed interaction tasks. Users control virtual molecules and subsets that are registered to physical markers and try to combine those by selecting and then confirming a specific bond. When a desired bond has been selected, the user needs a way to confirm that bond without letting an atom go out of position. We developed and investigated two separate methods of confirming a selected bond when both hands are already doing a two-handed symmetric interaction task. The first method is a waiting method and the second method is a back&forth motion gesture. We evaluated the two methods in a user study, showing that the first technique, holding still, outperforms the other technique.

Because the resolution of see-through displays is lower than the resolution of the human eye, perception of AR schemes is complicated in large distances. To discover, how design issues of perception correlate with presentation in large distances, we developed three different variants of arrow-based route guidance systems. With a large-scale head-up display having a large focal depth, we tested the variants under different conditions on perception and interpretability.

Most car accidents occur due to longitudinal collisions or lane departure. We assume that the number of such accidents can be reduced, if the driver knows more precisely, where the car is heading and at which distance it can stop. To provide drivers with this kind of anticipation, we have developed two augmented reality based visualization schemes for longitudinal and lateral driver assistance in the head-up display (HUD) of cars. One presentation scheme indicates the braking distance by a virtual bar on the road. The second scheme adds the visualization of a drive-path between the car and the bar, zoning the entire region that the car will pass before coming to a complete halt. We have tested both schemes in a driving simulator in comparison to a baseline without visual assistance. Our results show, among other findings, that the bar is preferred, that it supports driving performance and that it does not increase mental workload.

Spatial sensor systems in cars are gaining more and more importance. Such sensor systems are the foundation of future safety systems, such as automatic emergency brakes, as well as for interactive driver assistance systems. We have developed a system that can visualize such spatial sensor data. Two environments are supported: a laboratory setup for off-line experience and a car setup that enables live experience of spatially aligned laser scanner and video data in real traffic. We have used two visualization devices, a video see-through LCD flat panel (TFT) and an optical see-through head-mounted display (HMD) in both setups. For the laboratory setup, a back-projection table has been integrated as well. To present data in correct spatial alignment, we have installed tracking systems in both environments. Visualization schemes for spatial sensor data and for geometric models that outline recognized objects have been developed. We report on our system and discuss experiences from the development and realization phases. The system is not intended to be used as a component of real driver assistance systems. Rather, it can bridge the gap between human machine interface (HMI) designers and sensing engineers during the development phase. Furthermore, it can be both a debugging tool for the realization of environmental perception systems and an experimental platform for the design of presentation schemes for upcoming driver assistance systems.

In cars, augmented reality is becoming an interesting means to enhance active safety in the driving task. Guiding a driver's attention to an imminent danger somewhere around the car is a potential application. In a research project with the automotive industry, we are exploring different approaches towards alerting drivers to such dangers. First results were presented last year. We have extended two of these approaches. One uses AR to visualize the source of danger in the driver's frame of reference while the other one presents information in a bird's eye schematic map. Our extensions were the incorporation of a real head-up display, improved visual perception and acoustic support. Both schemes were evaluated both with and without 3D encoded sound. This paper reports on a user test in which 24 participants provided objective and subjective measurements. The results indicate that the AR-based three-dimensional presentation scheme with and without sound support systematically outperforms the bird's eye schematic map.

With recent advances of head-up display technology in cars, augmented reality becomes interesting in supporting the driving task to guide a driver's attention. We have set up an experiment to compare two different approaches to inform the driver about dangerous situations around the car. One approach used AR to visualize the source of danger in the driver's frame of reference while the other one presented information in an egocentric frame of reference. Both approaches were evaluated in user tests.