Open-Source Assistive Robotics

Replicating the guide dog's pull through 3D spatial detection & haptics.

EchoSight is an open-source wearable system that empowers the visually impaired to navigate autonomously. By combining active-stereo depth vision, on-chip 3D spatial object detection, visual SLAM mapping, and a 5-point forehead tactile array, the system translates physical obstacles into intuitive directional guidance.

VPU INFERENCE
25–30 FPS
On-chip Myriad X
REFLEX LATENCY
< 40 ms
Spatial depth fusion
DEPTH SENSING
Active IR
Dot laser projector
TACTILE ARRAY
5x LRA
Forehead directional pull
What Makes EchoSight Novel?

Silent Tactile Guidance with Reflex-First Edge AI

1. Zero Audio Masking

Leaves the user's ears 100% unobstructed so critical ambient acoustic cues (traffic, footsteps, echoes) remain intact.

2. Sub-40ms Reflex Loop

High-bandwidth 3D depth and YOLO run entirely on-chip in the VPU, delivering instant tactile avoidance before hitting host compute.

3. Active IR in the Dark

Active infrared dot projector paints artificial texture onto smooth floors and pitch-black rooms where standard stereo cameras fail.