Abstract:
In this dissertation, we explored the usage of computational interaction design methods in the construction of human-centered tools. Computational design interaction methods involve applying learning and optimization to develop user interfaces. Effectively integrating computational tools into interactions can mitigate the effort needed to use these tools, possibly also by flattening their learning curve. These tools can create a seamless workflow for users by eliminating repetitive and/or routine tasks, thus allowing them to focus on the more creative aspects of their work. The novelty and variety of interaction modalities in emerging technologies like VR/AR bring forth fertile ground for exploring incorporating computational tools in designing interactions. AR/VR introduce additional interaction challenges such as maintaining environmental awareness and creating immersive experiences, that could be solved by using computational interaction design methods. We leveraged the human-centered design paradigm to develop several computational interaction tools: - A VR navigation tool that is adaptively displayed according to the user's need for navigation help, as determined by a gaze-based learning model. - A tool for automatically arranging visual elements in a virtual space by optimizing for the amount of visual attention they could receive. - An environment-aware mobile AR drawing tool that allows users to draw 2D strokes on a tablet that are projected and optimized into 3D curves on the environment geometry. - A character design tool that furnishes designers with the capability of exploring their cartoon character colorization designs with a generative auto-colorization model.
