Dignity shouldn't cost a subscription.
EchoSense began with a simple frustration: the best assistive technology is locked behind cloud subscriptions and prices that put it out of reach for the people who need it most. We set out to prove that genuinely helpful, deeply private AI could run on affordable, off-the-shelf hardware — and reach the communities that mainstream technology overlooked.
of conversation sent to the cloud
offline voice-assistant tools on Echo 1 — zero cloud
on-device obstacle detection (Hailo-10H)
passing on real wearable hardware
Eight phases, measured at every step
The Echo 1 wearable was built and verified phase by phase on real hardware — nothing was assumed, everything was benchmarked.
Hardware validation
Every sensor independently bench-tested and bring-up scripted.
Core device framework
Config, logging, event bus, supervisor, audio manager, self-check.
Tool-calling assistant
Push-to-talk → STT → intent router → tool registry → spoken reply.
Safety engine
Camera + LiDAR fusion, hazard evaluator, alert manager — 30 FPS.
Navigation engine
GPS, route state machine, clock-direction turn cues, rerouting.
Vision AI
Offline scene description, OCR and find-object.
Operations dashboard
Real-time WebSocket SPA, read-only and process-isolated.
Deployment
systemd autostart with supervisor + systemd two-layer restart.
Built by the EchoSense team
Khaled Eissa
Hardware & Safety Engine
Mostafa Khalid
Hardware & Safety Engine
Yousef Haytham
Mobile App & On-Device AI
Ahmed Khairallah
Mobile App & On-Device AI
Sohila Alnahhas
UI/UX Engineer
Haidy Ahmed
UI/UX Engineer
Want to collaborate?
We're always glad to hear from clinicians, accessibility advocates, researchers and the community.
Get in touch