Harvard researchers develop AI app to aid blind navigation

A team of engineers from Harvard University has developed an innovative AI-powered application called Mobilio to assist individuals who are blind or visually impaired in navigating their surroundings safely. The application, which utilizes the functionalities present on smartphones such as the camera, GPS, and motion sensors, offers a comprehensive solution for these users to move around independently and avoid potential hazards.

To gather insights for the app's development, researchers conducted a survey among over 100 blind and visually impaired individuals. The findings of this survey were instrumental in designing Mobilio, which has shown promising results in initial tests involving 14 participants. When used in conjunction with a white cane, the app enabled users to complete outdoor routes approximately 13% faster than using Google Maps with a cane. Furthermore, Mobilio helped reduce collisions with obstacles on indoor routes by around 41%.

Mobilio operates by continuously analyzing the user's surroundings through the phone's camera while simultaneously utilizing GPS and motion sensors to determine the user's location and direction. The app then provides audio instructions that guide the user's movements and adapt to the user's responses. Its computer vision model is specifically trained from a pedestrian's perspective, enabling it to recognize essential elements such as sidewalks, crosswalks, and pedestrian pathways.

This capability allows Mobilio to offer continuous navigation assistance, rather than merely detecting nearby obstacles.