Haptic Vision provides a wearable vest that combines a stereo depth‑sensing camera, on‑device AI narration, and a three‑zone haptic belt to give blind and low‑vision users real‑time spatial awareness.
Funding
Funding not disclosed
Founders
Product
Problem
Visually impaired individuals often lack reliable, real-time spatial awareness, making navigation in unfamiliar or dynamic environments hazardous and limiting independence.
Solution
Haptic Vision offers a wearable vest that fuses depth-sensing stereo vision, on-device AI narration, and multi-zone haptic feedback to deliver immediate, proportional cues about surrounding obstacles and people. The ZED 2i camera captures 1080p depth data up to 8 m, while an NVIDIA Jetson Orin NX processes this information locally with a Google Gemma 4 model, generating natural‑language scene descriptions without internet connectivity. Three vibrating motors on the belt provide intensity‑scaled haptic alerts aligned to left, center, and right zones, keeping the user centered and aware of hazards. Spoken alerts and facial‑recognition announcements further enhance safety, and the system can record routes to provide turn‑by‑turn navigation using combined audio and haptic cues. All processing occurs on-device, ensuring low latency and preserving user privacy.
Target Audience
The primary users are blind and low‑vision individuals seeking independent mobility, as well as assistive technology providers and rehabilitation centers that equip users with advanced navigation aids.
Features
- Depth perception via ZED 2i stereo camera with 1080p capture and 8 m range, functional in any lighting condition
- On‑device AI scene narration powered by Google Gemma 4, delivering natural‑language environment descriptions without internet
- Three‑zone haptic belt with proportional vibration intensity (Weber‑Fechner scaling) for obstacle proximity and directional guidance
- Real‑time facial recognition that announces identified individuals by name and direction
- Automatic route recording and waypoint‑based turn‑by‑turn navigation using combined haptic and audio cues
- Dynamic warning range that adapts to walking speed and heading lock via magnetometer for drift correction
- Fall detection using barometric pressure and IMU sensors
- Fully local hardware stack (NVIDIA Jetson Orin NX, Arduino Uno, ZED 2i camera, Google Gemma 4) for low latency and privacy