Xu Zhen
From Graphic to Installation: Multi-factors Analysis to Realization Process of Interactive Design
Abstract: Advanced fabrication techniques have grown in sophistication over the last decade, vastly extending the scope of structures and materials that can be manufactured. While providing new opportunities for personalized fabrication, product design, engineering, architecture, art, and science, the potential impact of these techniques is tightly coupled with the availability of efficient computational methods for design.
In this talk, I will describe our recent efforts toward transforming design workflows with computational and data-driven methods. I will first introduce a generic optimization approach based on the extended finite element method, which allows for the optimization of a wide range of design objectives directly on parameterized 3D CAD models. Leveraging optimization-based design and a tailored data-driven model of the materials' responses, I will then introduce novel approaches for interactive shape exploration and demonstrate its applicability to designing cold-bent glass facades and deployable structures. Furthermore, I will show how insights from geometry can be used to derive an intuitive and rigorous characterization of the design space of planar elastic rods, which can be shaped into intriguing curved elements.
Finally, I will reflect on the successes and the challenges of algorithms and artificial intelligence as tools for the future of design and discuss research opportunities in this area.
Nowadays, computer became able to create several types of design automatically based on models, and media has praised it by calling it as AI design. When it is easy to make the models, this is a good auto-design method. However, there are many real-world design tasks which computational models are not easily made, while human can can draw their design images in their mind. Creating computer graphics that I like, tuning hearing-aid which sounds are easy to hear for me, controlling robot's behavior elegantly, and other are example design tasks. Key issue for this kind of design is human factors. AI approach models human functions, such as handling knowledge, learning, reasoning, and others, and uses the models in computer instead of humans. Contrastively, design based on Humanized Computational Intelligence emphasizes the importance of good cooperation between humans and computer; precisely speaking, cooperation of their different capabilities. In this talk, firstly, we emphasizes the importance of human factors with some examples. Secondly, we introduce interactive evolutionary computation (IEC) as a tool for Humanized Computational Intelligence. We show several concrete design tasks using human-computer cooperation through IEC in artistic design, engineering design, and others. Finally, we shortly introduce new research direction: IEC for human science. We can analyze human characteristics or find new knowledge on humans by analyzing a system optimized using IEC and an IEC user. Those who are interested in IEC can download about 100 slides and a tutorial paper from http://www.design.kyushu-u.ac.jp/~takagi/TAKAGI/downloadablefile.html#IEC.