Inspiration

Navigating unfamiliar indoor spaces can be stressful and unsafe for people who are visually impaired — especially in environments like universities, offices, and public buildings. To help others understand this challenge, we asked people to imagine being in an unfamiliar building during a fire alarm, unsure where the exits or obstacles are. For the 2.2 billion people worldwide living with visual impairments, this uncertainty is a daily reality. GuideMate was inspired by the need to replace that anxiety with clarity, confidence, and safety.

What it does

GuideMate is a web-based indoor navigation assistant that provides real-time voice guidance for visually impaired users. Using natural voice interaction, users can ask GuideMate how to get to a destination — such as the nearest exit — and receive clear, step-by-step audio directions. Along the way, GuideMate detects obstacles, recognizes landmarks, and continuously adapts as the user moves through the environment.

Note: GuideMate's functionality is localized to the second floor of University College as a MVP.

How we built it

GuideMate is powered by computer vision that continuously detects objects and identifies key landmarks within indoor spaces. We integrated ElevenLabs for both speech-to-speech and text-to-speech, allowing users to speak naturally and receive clear, real-time voice guidance in return. The application is built using TypeScript and hosted on the web with Cloudflare, making it fast, reliable, and easily accessible from any device without requiring a native app.

Challenges we ran into

One of our biggest challenges was designing guidance that was simple, clear, and non-overwhelming while still being informative enough to ensure user safety. Creating the navigation was difficult because GPS supported systems are limited in indoor spaces. Balancing real-time computer vision processing with low latency voice feedback was also difficult, especially in a web-based environment. Additionally, ensuring accessibility and usability across different unfamiliar indoor layouts required careful design decisions.

Accomplishments that we're proud of

We’re proud to have built an accessible, web-based solution that combines computer vision and natural voice interaction to support indoor navigation. Creating a system that adapts continuously as the user moves — while remaining intuitive and easy to use — was a major milestone. Most importantly, GuideMate demonstrates how technology can meaningfully improve independence and safety for visually impaired users.

What we learned

Through building GuideMate, we learned the importance of designing with accessibility as a first-class priority rather than an afterthought. We gained experience integrating real-time computer vision with speech technologies and learned how critical clarity, pacing, and user trust are when designing assistive tools.

What's next for GuideMate

Next, we plan to introduce personalized guidance settings, allowing users to customize instruction style, pacing, and alert preferences. We’re also working on emergency-specific modes, such as automatically guiding users to the nearest exit when a fire alarm is detected. Finally, we aim to expand the number of indoor spaces GuideMate supports, making safe indoor navigation more accessible everywhere.