Skip to main content
DwellTransit
Live now

Real-time arrivals university riders can actually trust.

Dwell is one platform for university transit. It rebuilds every bus prediction from raw GPS, powers the rider and driver apps, and proves its accuracy out in the open.

Every prediction is measured against established accuracy benchmarks, and the results are published openly. See the live accuracy →

The whole lifecycle, end-to-end

Dwell owns every stage from the raw GPS to the rider's phone, and each stage makes the next one better.

  1. 01
    Capture raw GPS

    Every prediction starts from one source of truth: the raw GPS each bus reports. Dwell works straight from where the vehicles actually are.

    Raw GPS Multi-source ingest Google Cloud
  2. 02
    Snap to the route

    The Snapper map-matches each GPS point onto the route's real shape, so Dwell knows exactly where a bus is along its path, and which stop comes next.

    Python Map-matching GTFS shapes
  3. 03
    Store the history

    Every snapped observation is saved, building per-stop and per-segment timing from millions of past trips, by time of day, day of week, and service period.

    BigQuery Per-segment timing
  4. 04
    Predict & publish live GTFS feeds

    Live position meets historical timing to produce arrival predictions, published as standards-based GTFS-Realtime: VehiclePositions and TripUpdates.

    Python GTFS-Realtime Cloudflare edge
  5. 05
    Rider app

    One native app for every campus. Riders choose their university; each campus sets its own color and logo. Live map, arrivals, and routes in riders' pockets.

    Swift Kotlin Campus selector
  6. 06
    Driver app

    A companion app for operators. Drivers see their route, schedule adherence, and spacing in real time, which keeps service steady on the road.

    Swift Kotlin On-route tools
  7. 07
    Measure accuracy against benchmarks

    Every prediction is compared with what actually happened, scored against established transit-accuracy benchmarks, and published openly, so anyone can check Dwell's work.

    BigQuery Established benchmarks
  8. …and it compounds

    Every trip adds to the history and the accuracy record, sharpening the next prediction. Dwell gets more accurate the longer it runs on your campus.

Built for the academic calendar

A campus in July moves nothing like a campus in October.

Dwell learns separate timing models for each service period (summer, semester, and breaks), so arrivals reflect how the campus actually moves right now instead of a year-round average.

What your university gets

Arrivals riders trust

Predictions rebuilt from raw GPS and checked against established accuracy benchmarks, so the times on the screen hold up.

One app, your campus

Riders pick their university and get your colors and logo. No separate app to build or maintain.

Works with the maps riders already use

Standards-based GTFS and GTFS-Realtime feeds plug into the transit apps students already have.

Open accuracy

A public accuracy page so riders and administrators can see how Dwell is performing, anytime.

Bring Dwell to your university

Tell us about your campus and routes. We'll show you how Dwell performs for your system and get your riders an app and feeds they can trust.