Dissertation Defense: Austin Wilson, "Development of Occasion-Capable Optical See-Through Head-Mounted Displays for Augmented Reality"

    Monday, January 10, 2022 - 12:00pm

    Join Zoom Meeting


    Password: 3dvis


    Conventional optical see-through head-mounted displays (OST-HMDs) for augmented reality (AR) applications lack the ability to properly render correct light-blocking behaviors between digital and physical objects in space. When this ability, known as mutual occlusion, is achieved, a solid virtual object should appear to be fully opaque and occlude a real object located behind it. Likewise, a real object located in the foreground should naturally occlude the view of a virtual object located behind it. Most OST-HMDs are unable to achieve this mutual occlusion because they typically rely on a simple optical combiner such as a diffractive grating or a beamsplitter to uniformly combine the light of the real world with virtual objects. Consequently, virtual content rendered by these displays appears semi-transparent and floating or as an indistinct blend of both real and virtual objects.

    Lack of mutual occlusion capability in state-of-the-art OST-HMDs can cause severe misjudgments, such as incorrect color registration, degraded image contrast, and object placement disparity.  Outdoor and bright environments can also present a major issue for optical see-through AR displays to correctly render color and contrast because background light often washes out text and blends colors. This can compromise the user’s interpretation of the virtual content, rendering color-specific interfaces useless, and can become problematic when color is especially important, such as in military applications or medical visualizations. 

    This dissertation discusses the complex challenges of rendering mutual occlusion in OST-HMDs and presents three new optical architectures for enabling a compact, high performance, occlusion-capable OST-HMD (OCOST-HMD). The optical design and optimization of each novel architecture are detailed along with a comprehensive performance analysis of each binocular prototype system. For each optical system, a contrast ratio greater than 100:1 for the occlusion module was achieved, significantly exceeding that of conventional see-through HMDs. Each prototype also maintained its performance in bright environments (> 350 cd/m2), without loss of contrast or color fidelity to the virtual image. Additionally, this work includes an analysis of how occlusion affects color and contrast as it pertains to the human visual system in a conventional HMD system versus an OCOST-HMD system and details various studies that blend light from real-world backgrounds with virtual light produced by the OCOST-HMD systems. The results are consistent with the demonstrated optical performance of the custom-prototyped systems.