Français Anglais
Accueil Annuaire Plan du site
Home > Research results > Dissertations & habilitations
Research results
Ph.D de

Group : Large-scale Heterogeneous DAta and Knowledge

Algorithmique robuste pour l'accès aux données en présence d'ontologies

Starts on 01/10/2013
Advisor : GOASDOUE, François

Funding : contrat doctoral UPS
Affiliation : Université Paris-Sud
Laboratory : LRI LaHDAK

Defended on 29/09/2016, committee :

Research activities :

Abstract :

Ph.D. dissertations & Faculty habilitations


Industrial design reviews benefit from emerging interactive technologies to become more Realistic, Immersive and Collaborative. The industrial product review is a vital process for project members to assess aesthetic properties, user satisfaction, and technical feasibility of the product before physical prototyping. Particularly, reviewing digital mock-ups in a realistic context with one-to-one scale enriches their reviewing experience, and a shared workspace facilitates discussions across distinct experts. However, modifying the product data during the reviews is still challenging. Although research in the integration of Virtual Reality (VR) and Computer-Aide Design (CAD) has proposed a range of systems for design reviews, only a very few of them support direct modifications of original CAD data within immersive systems. Engineers, therefore, need to apply post modifications on the CAD data from a workstation to adjust the design based on discussions during the design reviews. I argue that current distinct processes of design reviews: discussion and design adjustment should merge—so thus it could reduce the iterations, facilitate discussions and empower all users to directly apply modifications on CAD data. In this dissertation, I propose a new industrial design reviewing paradigm in which project members can adjust and compare a final design of the product within interactive systems. Considering various traits of experts involved in the review process, the interactive systems should be configurable to meet their requirements, and collaborative within or across the heterogeneous systems. As the accessibility of native CAD data is the cornerstone for the new design review process, I first designed a back-end server, namely VR-CAD Server, which can update the CAD data with an embedded CAD engine and transmit it to the interactive systems. I implemented VR-CAD Server in a distributed network architecture that makes the system highly configurable to support heterogeneous systems and multi-user interactions between remote locations. This structure is a basis of "modifiable" project reviews, which adds a modification capability on top of the current design reviews. Based on the distributed architecture, I explored interaction techniques for novices to modify parametric CAD data in large interactive systems. Since targeted users include non-CAD experts, interactions with the 3D-CAD model should be straightforward to learn, i.e. without the interaction on parameters, over heterogeneous systems. The choice of interactive systems differs according to the expertise and their purpose: e.g. ergonomists are likely to choose a 3D environment to assess the spatial feeling, whereas project managers might prefer discussing with a large format 2D display. I therefore designed two interaction techniques on 3D and 2D interactive systems to cover diverse reviewing scenarios, and conducted user studies for each: a Cave Automatic Virtual Environment (CAVE) system and a wall-sized display. For the former case, I designed a 3D interaction technique on parameter modifications of the CAD data, namely ShapeGuide. With ShapeGuide, users can implicitly manipulate parametric constraints of a CAD part with a co-localized shape-based interaction. This technique prepares a set of shape variations from an original design at run-time to guide users’ hand gestures. To stabilize the gestures in 3D space, I also tested force feedback during manipulations. I performed a controlled experiment to evaluate how ShapeGuide affects a CAD data modification task in comparison to a standard one- dimensional scroll technique. Results of the experiment demonstrate that ShapeGuide is significantly faster, more efficient and preferred by the users than the scroll technique. As for the wall-sized display, I designed ShapeCompare, an interaction, and visualization technique in which users can modify and compare multiple designs of CAD models on a large space with a touch interaction. The technique of ShapeCompare is based on ShapeGuide, which generates and presents multiple design alternatives deviated from an initial CAD model on the wall. I performed two experiments to evaluate how a large number of design alternatives displayed on a wall-sized display affects on collaborations between different experts, and how it helps their design explorations. Results of the experiments showed that with ShapeCompare, paired participants finished the collaborative reviewing task faster, preferred and found it more helpful in communications across pairs. Lastly, I conclude with an illustration of future collaborative design review scenario across heterogeneous systems. I implemented a proof-of-concept between a CAVE system and a Wall-sized display in which both users can review and modify the CAD model in a distant location.