The problem TASE solves
TASE solves the everyday independence gap faced by visually impaired people: moving safely through real, chaotic environments and understanding what’s around them without relying on other people or unreliable infrastructure.
A cane can detect obstacles only when they’re very close and provides no awareness of overhanging objects, stairs, moving obstacles, or complex indoor layouts; meanwhile most “smart” assistive devices either become bulky/expensive (sonar/LiDAR-based) or depend on cloud connectivity that introduces latency, recurring cost, and privacy risk, especially impractical in many Indian settings with inconsistent internet access and crowded, noisy streets.
The core problem is not just obstacle detection, but real-time, context-aware guidance and interaction: knowing where an obstacle is, how far it is, and being able to ask “what is this?” or “read this” instantly.
TASE addresses this by turning visual information into intuitive spatial audio cues and enabling offline visual Q&A, so users can navigate and understand their environment continuously, affordably, and privately.
Challenges we ran into
- First and foremost was to test if the solution we're building actually helps the visually impaired. For our research to build the perfect solution, we needed first hand reviews from people actually living those conditions.
- Converting depth into surround sound cues required heavy tuning because frequent alerts can overwhelm the user, and multiple obstacles create audio clutter unless you prioritize the most critical one.
- Running fully offline on a Raspberry Pi 3 B+ exposed hard limits on latency, frame rate, heat, and battery life, and even small delays reduce safety and user trust.
- Wearability and integration were nontrivial, camera angle and mounting stability affected perception, audio delivery had to preserve ambient awareness, and comfort and durability mattered as much as the model.