Bono

Drishti (दृष्टि) is a safety-first AI walking companion for blind and low-vision people. A phone camera worn on a chest strap or lanyard feeds frames to two detection tiers: an on-device detector that catches approaching people, vehicles and furniture, and Meta's Muse Spark model, which identifies stairs, corners and other hazards. A priority arbiter decides what is spoken, so a hazard alert always interrupts lower-priority speech. It speaks in Hindi or English, tells the user when the camera is blocked, and includes a user-initiated emergency flow: three taps, a 10-second cancelable countdown, then a prefilled SMS and a one-tap 112 link. It complements a cane or guide dog and does not replace one. This is an MVP on a phone. The design is glasses-ready, and the target device is Ray-Ban Meta glasses, which are not integrated yet. Next steps with Meta's stack (planned, not yet built): SAM 3.1 to segment the walkable path and show which side has clear space; Muse Voice transcription for hands-free voice commands in Hindi and English; the Meta Wearables Device Access Toolkit to take the camera stream from Ray-Ban Meta glasses and play guidance through their speakers; and further Muse Spark models for scene understanding and nearby-place guidance.