Higher Degree Research Projects

Wayfinding and navigation are essential aspects of independent living that heavily rely on the sense of vision. Walking in a complex building requires knowing one’s exact location to find a suitable path to the desired destination, avoiding obstacles and monitoring one’s orientation and movement along the route. People who do not have access to sight-dependent information, such as that provided by signage, maps and environmental cues, can encounter challenges in achieving these tasks independently. They can rely on assistance from others or maintain their independence by using assistive technologies and the resources provided by smart environments. Several solutions have adapted technological innovations to combat navigation in an indoor environment over the last few years. However, there remains a significant lack of a complete solution to aid the navigation requirements of visually impaired (VI) people. The use of a single technology cannot provide a solution to fulfil all the navigation difficulties faced. A hybrid solution using Internet of Things (IoT) devices and deep learning techniques to discern the patterns of an indoor environment may help VI people gain confidence to travel independently.

The work proposes a novel framework, Indoor-Nav, to provide a VI-friendly path to avoid obstacles and predict the user’s position. The components include Ortho-PATH, Blue Dot for VI People (BVIP), and a deep learning-based indoor positioning model. The work establishes a novel collision-free pathfinding algorithm, Orth-PATH, to generate a VI-friendly path via sensing a grid-based indoor space. Further, to ensure correct movement, with the use of beacons and a smartphone, BVIP monitors the movements and relative position of the moving user. In dark areas without external devices, the research tests the feasibility of using sensory information from a smartphone with a pre-trained regression-based deep learning model to predict the user’s absolute position. This work has kept low-cost solution and enhancement using latest technology for those who need it most as primary requirements. The approaches and framework of the work contributes   to analyse the capability of IoT implementations and deep learning techniques to enhance and improve the indoor journey of VI people.