Augmented Reality Navigation Assistance for Visually Impaired

Moving around independently is still difficult for people with vision loss because they often need help from others or face barriers in accessibility remains a major challenge and dependence on external assistance. This study proposes a smartphone-based AR navigation system that can detect obstacles and give real-time guidance. The goal is to help visually impaired individuals move more safely and confidently, without any extra devices. Our system uses computer vision methods to identify obstacles and detect safe pathways, while providing audio instructions to assist users during navigation. The COCO dataset is employed for training and testing object recognition, and the models are optimized with lightweight algorithms to achieve fast, real-time execution on smartphones. The design focuses on minimizing latency, reducing computational load, and maintaining high accuracy for indoor and outdoor environments. Preliminary results indicate that the system can accurately identify obstacles such as stairs, doors, and vehicles, while delivering clear audio instructions to assist navigation. This study shows that AR-based guidance is effective even without the use of specialized wearable devices, making the solution more affordable and accessible. In conclusion, this study offers an innovative AR solution that combines familiar assistive tools with cutting-edge navigation technology, giving visually impaired individuals more confidence and freedom in their daily movements.