NSF Convergence Accelerator Track H: Visit Unknown Places Confidently: Mapping for Access BuiLt Environments (MABLE)

Lehigh University, Bethlehem PA

Investigators

Abstract

Wayfinding and navigation within unfamiliar indoor built environments are challenging propositions for many people, including those with disabilities. Blind or low-vision users struggle due to the lack of user-friendly signage and their inability to create a mental map of the space. Individuals with mobility impairments find it difficult to identify routes and know upfront about potential challenges they may encounter. Individuals with cognitive impairments struggle to create or recollect mental maps they may have made. Such challenges in navigating unfamiliar spaces is a barrier to participation in social and economic life. Technology that helps to navigate spaces reduces barriers therefore, increasing the range of regular activities conducted independently, integrating into the workforce, and increasing productivity. The project has the long-term goal of improving access within and around indoor built environments through the creation of MABLE (Mapping for Access in BuiLt Environments). MABLE will provide digital maps of indoor environments with an interface for assessing, planning, and navigating within them based on the affordances and capabilities of the user. It will also permit map augmentation by users based on their experiences and observations. Achieving both scalability and ease-of-use with indoor maps is a challenging proposition. Current companies focus on mapping, that can scale, tend to prioritize features or offerings that do not cover the richness needed for navigation. Other companies focus on scalable offerings with reduced labor demands and deep learning enabled image processing. Others that rely on user contributions for scalability cannot guarantee richness and completeness of information, and do not integrate localization tools that can enable real-time turn-by-turn navigation and exploration using contextual information. To achieve both scalability and ease-of-use, MABLE proposes to leverage advances in artificial intelligence (AI), building modeling, robotics, augmented reality (AR) and virtual reality (VR) visual scene reasoning, and low-power consumer electronics. MABLE will extract most usable-related information directly from floor plans using deep-learning augmented image-processing algorithms, and any missing information can be augmented through robot mapping and surveying, and stakeholder and user contributions. Further, by quantifying the localization performance and costs of an array of indoor localization technologies, the MABLE product will create custom localization deployments that work best for a user, encouraging greater adoption. The project will create new knowledge in the areas of collection, processing, and evaluation of information from built environments. In addition to a long-lasting model, the project will also create new frameworks for quantifying economic benefits from built environments encompassing perspectives of future economic growth potential, cost savings, and return on investments. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.