Plausible multisensory modeling of 3D deformable objects for interactive applications

在市场上引入3D使能技术，例如3D显示器、3D眼镜和用于与虚拟环境交互的各种手段（手套、触觉反馈设备等）。对研究界施加了越来越大的压力，以开发3D内容的支持技术。为了充分发挥其潜力，交互式环境需要超越软件生成对象的建模，引入基于真实的测量的3D对象的合理表示。现实世界中的物体所表现出的丰富的交互行为不仅取决于物体的外观，还取决于它们的属性：它们有多硬，它们的表面在触摸时的感觉和声音，或者它们在接触时如何变形。基于真实的测量的对象的复杂行为的交互式模型的发展没有跟随娱乐和计算机游戏产业中实现的视觉和音频真实性的快速进步。为了应对这一挑战，该研究计划的长期目标是通过基于真实的多传感器测量的合理3D对象模型，沿着其集成，表示和解释，为需要更高真实感的应用开发增强的传感和建模技术。从这个长期目标涉及的众多方面来看，这个原始的研究计划集中在三个主要方面：（1）视觉，听觉和触觉测量的获取和集成到逼真的，连贯的3D模型中;（2）渲染这些对象的复杂行为;（3）在特定应用的背景下集成和评估多感官对象模型。更具体地说，人类视觉和触觉传感方法、使用各种传感器的触觉和视觉采样策略、视觉算法和计算智能技术将被调整和改进，以用于真实的测量。它们将被有利地结合起来，用于基于多感官视觉、触觉和音频数据的新颖对象模型的制定。原始的计算智能方法将被提出，以允许有效地利用各种来源的信息的对象的行为和无缝合并的多感官模型在交互式环境中。在初始阶段，将建立一个专用界面，用于可视化和评价所开发的模型。接下来，模型将在洞穴自动虚拟环境（CAVE）中进行验证，在那里，小的缺陷会立即被发现。模型最终将被集成到两个实际应用中，一个用于恐惧症治疗，另一个用于安全的基于模型的可变形对象的对象操作。拟议的计划将有助于建立下一代的交互式应用程序，将依赖于似是而非的3D内容从真实的测量，而不是硬编码，近似计算机生成的图形模型所需的理论和实践知识。这一研究领域的进步将对需要更高精度和更高真实感的交互式应用产生重大影响。除了上述两种应用之外，还包括用于培训的交互式环境，用于精确敏感的工业CAD流程的交互式虚拟原型设计，智能制造以及机器人装配，远程机器人，医疗和辅助机器人等灵巧操作应用。这项研究工作将支持培训具有尖端方法和技术的高素质专家，这些专家是加拿大强大的多媒体，游戏和虚拟现实行业的需求。

The introduction on the market of 3D enabled technologies, such as 3D displays, 3D glasses and of various means for interaction with virtual environments (gloves, tactile feedback equipment, etc.) places an increasing pressure on the research community to develop enabling technologies for 3D contents. To fully reach their potential, interactive environments need to go beyond the modeling of software generated objects by introducing plausible representations of 3D objects based on real measurements. The rich interaction behavior exhibited by real-world objects depends not only on how objects look, but also on their properties: how stiff they are, how their surface feels and sounds when touched, or how they deform on contact. The development of interactive models for the complex behavior of objects based on real measurements has not followed the rapid progress of visual and audio realism achieved in the entertainment and computer gaming industry. Rising up to this challenge, the long-term goal of this research program is to develop enhanced sensing and modeling techniques for applications that require higher realism by means of plausible 3D object models based on real multisensory measurements, along with their integration, representation and interpretation. From the multitude of aspects involved in this long-term objective, this original research program focuses on three major aspects: (1) the acquisition and integration of visual, audio and tactile measurements into realistic, coherent 3D models; (2) the rendering of the complex behavior of such objects; and (3) the integration and evaluation of multisensory object models in the context of specific applications. More specifically, human visual and tactile sensing approaches, tactile and visual sampling strategies using various sensors, vision algorithms and computational intelligence techniques will be adapted and refined for use on real measurements. They will be advantageously combined for the formulation of novel object models based on multisensory visual, tactile and audio data. Original computational intelligence methods will be proposed to allow for an efficient use of various sources of information on the object behavior and for the seamless merge of multisensory models in interactive environments. In an initial stage, a dedicated interface will be implemented for the visualization and evaluation of the developed models. Next, the models will be validated in a Cave Automatic Virtual Environment (CAVE) where small imperfections become immediately perceptible. Models will finally be integrated into two practical applications, one for phobia treatment and the other one for safe model-based object manipulation of deformable objects. The proposed program will contribute the theoretical and practical knowledge required to build the next generation of interactive applications that will rely on plausible 3D contents acquired from real measurements, instead of hard-coded, approximate computer-generated graphics models. Advancements in this area of research will have a significant impact on interactive applications that require higher accuracy and improved realism. These include, beyond the two applications above, interactive environments for training, interactive virtual prototyping for precision-sensitive industrial CAD-based processes, smart manufacturing as well as dexterous manipulation applications such as robotic assembly, tele-robotics, medical and assistive robotics. This research work will support the training of highly qualified specialists with cutting edge approaches and technologies, who are in demand for the strong multimedia, gaming, and virtual reality industry in Canada.