Rider copilot
Door-to-door options with confidence %, transfer difficulty, elevator alerts, and safer alternatives.
AccessibleAI Complete Trip Copilot
Person-centered trip planning with accessibility confidence scores, plain-language risk notes, and an agency barrier dashboard — built for the whole Complete Trip, not just the transit leg.
Agency intelligence
Heatmaps of trip-failure hotspots, recurring barrier patterns, and exportable planning reports.
Governed AI
Rules + retrieval-grounded explanations — not unsupervised “chat” that invents curb ramps.
Interactive Copilot Demo
Choose a rider profile and the AccessibleAI copilot will evaluate trip feasibility, accessibility risk, transfer complexity, disruption alerts, and safer alternatives using governed rules and sample transit data.
| Tab / profile | Sample trip | Complete Trip risk focus |
|---|---|---|
| Wheelchair user | North Park → UCSD Medical Center | Elevator access, transfer window, platform gap |
| Low vision | Gaslamp → Balboa Park | Landmarks, detours, tactile cues, street crossings |
| Cognitive support | Chula Vista → Downtown SD | Transfer complexity, fare steps, simpler alternatives |
| Agency dashboard | Regional rollup (illustrative) | Barrier hotspots, recurring failures, planning gaps |
Choose a rider profile — each scenario uses sample regional context, not live agency data:
Agency Barrier Dashboard Illustrative view · sample scenario rollup
47 failed scenario legs (30 days)
8 recurring barrier hotspots
3 transfer nodes flagged
Top barrier types (scenario rollup)
Grid heatmap — aggregated from scenario testing (illustrative)
- Illustrative recurring elevator outage pattern at Old Town transfer (wheelchair personas)
- Illustrative missing curb ramp cluster — University Ave & 30th (sample scenario data)
- Illustrative long transfer wait + steep ramp — Civic Center (cognitive-load scenarios)
Demo uses illustrative San Diego County scenarios and GTFS/GTFS-RT-style sample data. Phase I will evaluate integration with public transit schedule feeds, real-time alerts where available, pedestrian accessibility indicators, and governed accessibility rules. This is not live routing and does not imply agency endorsement or operational deployment.
Transit Schedule + Accessibility Layer
Phase I will integrate public transit schedule data using GTFS and GTFS-RT where available, combining route timing, transfer windows, service alerts, elevator/escalator disruptions, stop accessibility notes, and pedestrian access indicators into one accessibility confidence score.
| Data layer | Phase I use |
|---|---|
| GTFS schedules | route, stop, trip, and transfer timing using sample regional context |
| GTFS-RT-style alerts / GTFS-RT where available | Delays, disruptions, service alerts |
| Stop accessibility notes | Boarding/accessibility constraints |
| Pedestrian path data | Sidewalk, curb ramp, slope, crossing risk |
| Elevator/escalator status | Station access risk |
| User profile | Wheelchair, low vision, cognitive support |
Sample regional data context: MTS (San Diego Metropolitan Transit System) and NCTD (North County Transit District). Payment/fare integration and paratransit booking are Phase II+ capabilities requiring agency partnership.
Research-Informed Data Architecture
AccessibleAI builds on established transit data standards such as GTFS and GTFS Realtime, combines them with pedestrian accessibility indicators, and uses governed AI explanations to help riders and agencies understand not only which trip is fastest, but which trip is most feasible, reliable, and accessible.
| Layer | Purpose |
|---|---|
| GTFS Static | Routes, stops, trips, schedules, transfer timing |
| GTFS Realtime / GTFS-RT-style | Alerts, delays, trip updates, cancellations |
| Pedestrian Network | Sidewalks, crossings, curb ramps, slope, distance |
| Accessibility Rules | Wheelchair, low vision, cognitive support constraints |
| Confidence Score | Combines schedule reliability, transfer risk, access barriers, and user profile |
| Agency Dashboard | Shows recurring trip-failure points, barrier hotspots, and planning gaps |
Key takeaway
AccessibleAI is not “another trip planner.”
It is a governed accessibility confidence layer on top of public transit schedules, real-time alerts, pedestrian access data, and rider-specific needs.
Phase I System Architecture
Click any layer to see how data flows through the Phase I prototype. Illustrative architecture — sample demo integration in Phase I.
The Phase I architecture connects public transit schedule data, real-time alerts where available, pedestrian accessibility indicators, and rider profile constraints into a governed rule engine. The system produces accessibility confidence scores for riders and aggregated barrier intelligence for agencies.
Phase I — 6 months · $196,500
- Months 1–2 GTFS/GTFS-RT ingestion, accessibility rule engine, architecture
- Months 3–4 Rider copilot alpha, confidence scoring, dashboard v1
- Month 5 Usability testing (n=12) with disability community stakeholders
- Month 6 Feasibility report, Phase II roadmap, government demo