Developing a Self-Calibrating Navigation System to Address the Orientation Challenges faced by Older People with Visual Impairments

There are over 2 million older people with sight loss in the UK(AgeUK\_2015); globally, it's estimated to be 253 million(Ackland\_et\_al\_2017). As the population ages, this figure will increase as sight loss disproportionally impacts older people. People with sight loss benefit from the white cane to get around, often in conjunction with smartphone navigation apps. These can include accessibility features such as screen readers, colour filters, and dexterity tools. However, a long cane can only detect obstacles below knee-height and within approximately 1m of the user. Moreover, most smartphone interfaces are not developed with sufficient compatibility for these accessibility features, limiting their usefulness. This makes it particularly challenging for visually impaired individuals(VIIs) to navigate safely in busy urban areas, especially where cars, bicycles, and pedestrians are found in close proximity.

To address this challenge, WeWALK have developed a revolutionary "smart" device that screws onto a user's existing cane, upgrading a tool relied on by over 50 million people. Our cane detects obstacles at knee-to-head height and uniquely connects to a user's smartphone to provide audio-based navigation via integration with Google Maps, Moovit, and Microsoft services. Efforts to enhance WeWALK's indoor positioning by the project partners have uncovered that seamless wayfinding in all environments, including outdoors, is possible with situational awareness. This project will address this challenge, including the last-mile problem where individuals fail to locate their final destination or transport route, creating a fully autonomous, comprehensive, safety-critical, and accessible navigation system for older VIIs.

Recent assistive smartphone apps have included computer vision for object recognition in addition to environmental mapping and localisation through the concept of simultaneous localisation and mapping (SLAM). Their underlying principle can be very useful in solving WeWALK's navigation constraints; however, to date, they have not demonstrated adequate system performance(accuracy/reliability/usability) for use in a safety-critical application such as enabling a VII to navigate, avoiding hazards, around an urban environment.

We will address the technical/safety challenges that have prevented widespread uptake of computer vision-aided navigation including environmental mapping and overall situational awareness for older and VIIs. We will build on the existing WeWALK architecture with new hardware and software, incorporating spatial sensing and advanced data analytics, providing safety-critical functionality and usability frameworks. This project has the potential to play a vital role in ensuring that the rising numbers of visually impaired people have full access to the urban spaces where we live, work, and play.