Research · Digital Twin
The first digital twin of the Saint Paul Skyway.
spskyway.com isn’t a picture of the skyway — it’s a living model of it. The real geometry of the network, live transit at its doorsteps, and the actual movement of people inside it, updating right now. No one has built this for the skyway before. Here’s what’s in it, input by input.
- 60+
- Connected buildings
- 5 mi
- Indoor network
- 4
- Levels mapped
What “digital twin” means here
A twin is only as honest as its inputs. Here are ours.
The geometry is real
Every corridor traced across four levels — Upper, Skyway, Ground, Tunnel — and verified on foot, entrance by entrance. When a bridge closes, the model closes it.
The transit is live
Real Metro Transit NexTrip departures, served right inside the map. Not a schedule — what's actually arriving at the stops around you.
The movement is measured
Small counters — Bluetooth listeners and camera tripwires — measure real foot traffic in the network: counts and dwell, never identities. The twin doesn't guess how the skyway is used; it measures.
Live feature · 01
Turn-by-turn routing through five miles of indoor city
Ask the twin how to get from any door to any other and it answers with a real route — through the skyway when there is one, across levels when it has to. The routing engine runs on the network’s actual geometry — it knows which level you’re on and tells you when to change, from tunnel to street to skyway.
It works because the underlying graph is the real one — the same corridors, bridges, and entrances that got verified on foot.
Live feature · 02
The bus is part of the building
Downtown doesn’t end at the skyway door — it ends at the bus stop. So the twin carries live Metro Transit NexTrip departures for the stops around the network: what’s actually coming, in how many minutes, right inside the map.
Walk the skyway to the right exit and step out as your bus pulls up. That’s what a model of downtown is for.
Live feature · 03
The skyway’s first real foot-traffic signal
Nobody has ever really known how many people use the skyway — estimates, anecdotes, and arguments, for fifty years. We put small counters in the network as a working proof of concept. Below is the feed from the sensor at Steven D’s in Town Center — the same numbers we see, minutes old at most:
Two kinds of counters, one rule. Bluetooth listeners hear the randomized identifiers phones broadcast anyway — linked to a person only if that person explicitly opts in through the app. Camera tripwires detect a crossing and store a number — no faces, no identities, no footage kept. Raw data is purged once it’s been turned into counts. What survives is the only thing we wanted: how many, and for how long.
Live feature · 04
Questions the twin answers today
Because the model is live, it can be asked things a static map can’t. Not projections — readings.
How many people are near the sensor right now?
The feed above is answering this one live — readings minutes old, around the clock. One recent weekday at Steven D's ran from single-digit overnight lulls to 36 devices at once.
How long do people stay?
Measured, not assumed: the median visit near Steven D's holds about three minutes, and the longest tenth stretches past twelve. That's how you tell a corridor people cross from a place where they linger.
When does the corridor peak?
Every hour is on the record. The busiest recent hour at Steven D's logged roughly a hundred tripwire crossings — so programming, staffing, and cleaning can follow the real curve instead of a hunch.
Is anyone finding my business?
Every listing on the map counts its own views and direction requests, and owners read those numbers in their own dashboard. When the district asks businesses to back something, that's a receipt in every shop's hand.
That’s the honest scope today: live readings from a young sensor network, plus per-listing numbers for owners. The board-ready, cross-district reporting built on top of it is in active development — we’d rather show you two real sensors than a rendering of fifty.
Privacy · a design decision, not a disclaimer
A twin of the place. Never of the people.
Counts, not surveillance
No faces, no identities, no stored footage. Passive sensing keeps only counts — a device is ever connected to a person only when that person opts in through the app.
No account required
Browse the map, get directions, check departures — all without signing in or being profiled.
Nothing for sale
We don't sell data and we don't share it with advertisers. The numbers exist to make the skyway better, period.
The full policy is public: spskyway.com/privacy ↗
What’s next
The twin gets sharper from here.
More sensors
Two counters run today as a working proof of concept. The plan: coverage across the network's key corridors, so every block has a real signal.
More detail
Deeper building interiors, richer place data, more of the network's texture in the model — block by block, verified the same way the rest was.
Indoor positioning
GPS dies indoors. The iOS app in beta knows where you are inside the network anyway — the twin, in your pocket, aware of your position.
See your downtown run like this.
Book a working session, or just open the live map and click around. A pilot is short and reversible — live in weeks, scoped to a civic budget.