Trail.

Inspiration

Accessibility is usually treated as a mobility checklist; are there, ramps, elevators, suitable door widths? But for many people, including all of us on the team, exclusion isn’t just physical. We all struggle with sensory overload from time to time: loud hallways, crowded corridors, harsh lighting, etc.

We realized that most navigation tools only help people get to a building. They do nothing to help people move through it safely and comfortably, especially if they have mobility needs or sensory sensitivities. Trail understands both physical barriers and sensory environments as equally real obstacles.

What it does

Trail is an indoor accessibility and sensory-aware navigation engine. It helps users:

• Find routes that avoid physical barriers (stairs, narrow doors, broken elevators)
• Anticipate sensory-heavy areas (noise, crowds, lighting intensity)
• Receive calm, human-readable warnings along routes
• Get alternate paths when environments become overwhelming

How we built it

We built Trail as a unified accessibility graph system:

• Parsed indoor floor plan elements into structured nodes (stairs, elevators, hallways, rooms)
• Assigned physical accessibility attributes to each node
• Layered in time-based sensory heuristics (noise, crowd density, lighting risk), uses MongoDB to store map and crowdsourced data
• Used Gemini to interpret floor labels, generate route warnings, and summarize issue reports
• Uses Elevenlabs to add a human touch (aka a TTS reader)!

Challenges we ran into

• Balancing usefulness with a calm UX (too many warnings = overwhelm)
• Defining AI’s role so it enhances decisions without replacing deterministic routing
• Data storage and integrating as many APIs as we could!

Accomplishments that we're proud of

• Unified physical + sensory accessibility into a single system (rare in navigation)
• Designed an MVP that works without sensors using smart time/location modeling
• Built explainable AI outputs judges can easily understand
• Created something directly aligned with Canadian accessibility priorities (AODA + experiential inclusion)

What we learned

• Accessibility is not one-dimensional
• Sensory environments can be as exclusionary as stairs
• AI is most powerful when used to interpret and communicate
• Designing for calm is just as important as designing for function

What's next for Trail.

• Pilot with Canadian campuses and public institutions
• Replace heuristics with community-fed and real-time sensory reports
• Expand to hospitals, Service Canada offices, and high-traffic indoor spaces
• Build partnerships with accessibility offices and advocacy groups

Built With

• css
• elevenlabs
• figma
• github
• javascript
• node.js
• python
• react
• snowflake