Inspiration
Millions of visually impaired people face daily challenges navigating unfamiliar environments such as railway stations, streets, and public buildings. Simple tasks like detecting obstacles, reading signboards, or recognizing nearby people become difficult without assistance. We were inspired to create AIVA to make navigation safer and more independent using AI.
What it does
AIVA is an AI-powered navigation assistant that uses real-time computer vision and spatial audio to guide visually impaired users. The system detects obstacles such as vehicles, stairs, and walls, reads signboards using OCR, and provides directional audio feedback. It can also identify known faces with consent and guide users safely through unfamiliar spaces.
How we built it
The system uses a smartphone camera to capture real-time video frames. Computer vision models detect objects and obstacles, while OCR extracts text from signs. Gemini multimodal capabilities interpret the scene and generate navigation instructions. Spatial audio feedback is then delivered through earphones to guide the user intuitively.
Challenges we ran into
One challenge was ensuring real-time performance while processing video frames continuously. Another challenge was designing clear audio guidance without overwhelming the user. We optimized detection pipelines and simplified audio cues to create a more natural navigation experience.
Log in or sign up for Devpost to join the conversation.