Date of Award
Campus Access Dissertation
Doctor of Philosophy (PhD)
In the view of the industry, the modeling design process is an overwhelming, non-trivial, and time-consuming task. As a result, creating such a design can take days, months, even several years. In this thesis, we investigate how human designers use human-centric computational-aided modeling design tools to assist the modeling design processes. We investigate:
- how we utilize the perception data obtained from the virtual environments to develop novel computational approaches of the layout designs for human movement comfort, accessibility, and convenience.
- how we utilize the semantic data obtained from real environments to develop novel computational approaches of the sound source assignment for the immersive experience in virtual reality.
- how we utilize the suggested images obtained from a set of annotated images to develop novel computational approaches of gallery design for generating style-unified gallery walls in augmented reality.
We then tackle various challenging computational design problems within our novel approaches. In the visual perception approach, we devise agent-driven optimization methods for the road signs assignment by a given layout, and we devise an adaptive algorithm to design a gallery by a suggestion system. In the auditory perception work, we devise an automatic algorithm to enhance the immersive of the given static panorama images through realistic sound sources assignment.
Huang, Haikun, "AI-driven Computational Design Tools For Synthesizing Human-centric Design" (2020). Graduate Doctoral Dissertations. 583.