Low vision is a visual impairment that falls short of blindness but cannot be corrected by eyeglasses or contact lenses. According to the Center for Disease Control and Prevention (CDC), about 19 million Americans have low vision and the number is still rapidly growing with the overall aging of the population. Low vision includes many different conditions, such as central and peripheral vision loss, extreme light sensitivity, and blind spots, which have brought low vision people difficulties in their daily life. Unlike people who are completely blind, low vision people have residual vision and extensively use their vision in daily activities. However, prior research has mainly focused on providing only audio and tactile feedback for blind users, overlooking the residual vision that low vision people have. While current low vision aids (e.g., magnifier, CCTV) support basic vision enhancements, such as magnification and contrast enhancement, these enhancements often arbitrarily alter the user's full field of view without considering the user's context, such as the tasks, the environmental factors, and the user’s visual abilities. As a result, these low vision aids are not helpful or preferred by low vision users in many important daily activities, such as navigation and socializing. To address this gap in low vision accessibility, I sought to deeply explore low vision people's visual perception, and design and build intelligent augmented reality (AR) systems that provide direct visual augmentations to empower low vision people in various daily activities. Different from the conventional low vision aids, my AR systems provide semantic visual augmentations that are tailored to users' visual abilities and tasks. Specifically, my doctoral research focuses on three directions: First, I conducted both qualitative and quantitative studies to explore low vision people’s visual abilities and preferences on different AR platforms including both video see-through and optical see-through AR glasses, deriving design guidelines for AR assistive technologies for low vision. Second, I explored low vision people’s experiences and needs in different daily tasks, such as shopping and navigation, and designed intelligent AR systems with tailored visual augmentations to facilitate these tasks for low vision people. Finally, to enable easy control and interaction with the visual augmentations, I also designed accessible interaction techniques on AR glasses for people with low vision. Evaluations have showed that my AR systems effectively improved low vision people’s performance and experiences in various daily tasks. Based on my doctoral research and findings, I distill design considerations and discuss future research directions for low vision accessibility.