We present a study that compares finger-based direct interaction to controller-based ray interaction in a CAVE as well as in head-mounted displays. We focus on interaction tasks within reach of the users' arms and hands and explore various feedback methods including visual, pressure-based tactile and vibro-tactile feedback. Furthermore, we enhanced a precise finger tracking device with a direct pinch-detection mechanism to improve the robustness of grasp detection. Our results indicate that finger-based interaction is generally preferred if the functionality and ergonomics of manually manipulated virtual artifacts has to be assessed. However, controller-based interaction is often faster and more robust. In projection-based environments finger-based interaction almost reaches the task completion times and the subjective robustness of controller-based interaction if the grasping heuristics relies on our direct pinch detection. It also improves significantly by adding tactile feedback, while visual feedback proves sufficient in head-mounted displays. Our findings provide a guideline for the design of fine grain finger-based interfaces.

Natural Interaction in virtual environments is a key requirement for the virtual validation of functional aspects in automotive product development processes. Natural Interaction is the metaphor people encounter in reality: the direct manipulation of objects by their hands. To enable this kind of Natural Interaction, we propose a pseudophysical metaphor that is both plausible enough to provide realistic interaction and robust enough to meet the needs of industrial applications. Our analysis of the most common types of objects in typical automotive scenarios guided the development of a set of refined grasping heuristics to support robust finger-based interaction of multiple hands and users. The objects' behavior in reaction to the users' finger motions is based on pseudophysical simulations, which also take various types of constrained objects into account. In dealing with real-world scenarios, we had to introduce the concept of Normal Proxies, which extend objects with appropriate normals for improved grasp detection and grasp stability. An expert review revealed that our interaction metaphors allow for an intuitive and reliable assessment of several functionalities of objects found in a car interior. Follow-up user studies showed that overall task performance and usability are similar for CAVE and HMD environments. For larger objects and more gross manipulation, using the CAVE without employing a virtual hand representation is preferred, but for more fine-grained manipulation and smaller objects, the HMD turns out to be beneficial.

Natural Interaction in virtual environments is a key requirement for the virtual validation of functional aspects in automotive product development processes. Natural Interaction is the metaphor people encounter in reality: the direct manipulation of objects by their hands. To enable this kind of Natural Interaction, we propose a pseudo-physical metaphor that is both plausible enough to provide realistic interaction and robust enough to meet the needs of industrial applications. Our analysis of the most common types of objects in typical automotive scenarios guided the development of a set of refined grasping heuristics to support robust finger-based interaction of multiple hands and users. The objects' behavior in reaction to the users' finger motions is based on pseudo-physical simulations, which also take various types of constrained objects into account. In dealing with real-world scenarios, we had to introduce the concept of Normal Proxies, which extend objects with appropriate normals for improved grasp detection and grasp stability. An expert review revealed that our interaction metaphors allow for an intuitive and reliable assessment of several functionalities of objects found in a car interior.

Besides visual validations of virtual car models, immersive applications like a Virtual Seating Buck enable car designers and engineers to decide product related issues without building expensive hardware prototypes. For replacing real models, it is mandatory that decision makers can rely on VR-based findings. However, especially when using a Head Mounted Display, users complain about an unnatural perception of space. Such misperceptions have already been reported in literature where several evaluation methods have been proposed for researching possible causes. Unfortunately, most of the methods do not represent the scenarios usually found in the automotive industry, since they focus on too large distances of five to fifteen meters. In this paper, we present an evaluation scenario adapted to size and distance perception within the reach of the user. With this method, we analyzed our standard setups and found a systematic error that is lower than aberrations reported by earlier research work. Furthermore, we tried to mitigate perception errors by a Depth of Field Blur applied to the virtual images.

We are investigating the utility of a projection-based stereoscopic two-user system for applications in the automotive industry. In this paper we compare real-world pointing to pointing at virtual objects in a two-user scenario. In our study we investigated the following situation: One user points at an object while the second person has to guess the referenced object. Our results indicate that pointing at objects in a virtual world is much less precise than pointing in the real world even for a well calibrated stereoscopic two-user system. Pointing techniques like outlining an object or pointing from a distance produce more errors than the corresponding real-world techniques, but they are less error prone than interactions requiring to touch a virtual object. Our findings are a first step to qualify direct interaction techniques in a multi-user projection-based system.

We present a new tactile feedback system for finger-based interactions in immersive virtual reality applications that consists of shape memory alloy wires wrapped around tracked linger thimbles. The wires touch the inside of the finger tips and provide an impression when they are shortened. We use this system to communicate linger contacts with virtual objects in an application for usability and reachability studies of car interiors. Our experiments and an initial pilot study revealed that this type of feedback helps users to perform direct manipulation tasks with more reliability