AI-Based Real-Time Navigation and Obstacle Detection System for the Visually Impaired

Moving around independently is one of the hardest challenges for people with visual impairment or people with very little to no vision. Everyday tasks like detecting objects, finding routes become risky and difficult. There are traditional solutions like white cane and trained dogs but they are costly, limited ranged. The current review compares various assistive technology and other solutions. This paper presents an Android application that combines obstacle detection, outdoor navigation, emergency mechanism together in one smartphone application. The user speaks the destination, mobile camera detects objects and announces the detected object and approximate distance using a pre-trained YOLOv8n model, and it obtains a walking route from OpenStreetMap-based services and announces each turn. For safety, the app sends an SMS to the saved contacts along with the user’s location when an emergency is triggered and also when the battery drops below the threshold. This paper also describes the design and implementation of each feature, discusses the advantages and limitations of the application, and outline directions for future work.