A fixed timer gives the same green whether a lane is empty or backed up.
Timings follow what each lane actually looks like right now.
Maayong pag-abot - glad you’re here
Green goes where the traffic is.
ISTAP (Intelligent System for Traffic Analysis and Planning) is AI traffic signal management for Cagayan de Oro. It watches intersection feeds, counts lane queues, and recommends better light timings, so nobody sits at a red light for an empty road. An engineer approves every plan.
For city traffic teams, starting with Divisoria.
Try itSample data, not real readings
How busy is each street?
Real streets, made-up numbers. Move the sliders and watch the green time follow the queue.
Recommended green
22 sCorrales Ave
10 sT. Chavez St
Stays between 10 and 40 s in this sample, then waits for an engineer
- 🚌 Jeepneys
- 🚍 Buses
- 🛺 Motorelas
- 🏍️ Motorcycles
- 🚗 Cars
- 🚦 Queues
- 🌊 Standing water
- 🚧 Blockages
Divisoria
Where we want the pilot to start
3 languages
Bisaya, Tagalog and English
Every plan
Reviewed by a city engineer
0 tickets
No enforcement in the pilot
Sound familiar?
Maps show where traffic is slow. They can’t change the signal.
Since July 28, 2026, provincial PUVs end trips at terminals, and a league leader reported some commuters now take up to three rides.
A later update will give commuters ride and route guidance from the same live picture.
Floods and blockages cut off roads and bridges, and plans have to change quickly.
Planners can test closures in a sandbox before anything changes.
Watch it run
Same cars, same moment, two kinds of signal.
Corrales Ave green, 8 s left
- Average wait
- 51 s
- Waiting now
- 90
Corrales Ave green, 13 s left
- Average wait
- 41 s
- Waiting now
- 77
A simplified four-lane model of Corrales Ave x T. Chavez St running at 2.5× speed, with made-up traffic and a made-up neighborhood. It isn’t live or measured. Each approach takes its own green, the crosswalk on its right opens with it, and right turns can go on red after yielding to people crossing. Look for jeepneys, motorelas, motorcycles, hatchbacks, sedans, SUVs and pickups. A plus sign counts the vehicles queued out of view.
How it works
Seven steps from a camera to your commute.
Capture
Cameras at pilot intersections send video. The demo uses recorded or mock feeds.
Perceive
Gemini reads the frames and reports each lane: vehicle types, queues, blockages and standing water.
Decide
An optimizer proposes timings from that data and the situation, like rush hour or a flood.
Review
A city engineer sees the plan, the reason behind it and its predicted effect.
Act
Approved plans run in the simulation. Real signals need city authorization and come later.
Inform
In a later update, the same live picture will feed guidance for commuters.
Learn
Outcomes are compared with predictions, and the model is retuned.
Then it starts again.
Every run makes the next recommendation a little better informed.
Read the full walkthrough
Who it’s for
If you plan the road or ride it, this is for you.
City traffic teams
See each lane live, get signal plans with the reasons behind them, test them in a sandbox, and keep an approval log. More for city teams
Unsa nga intersection ang pinaka-traffic karon?
Why was green extended at this intersection?
Jeepney and PUV passengers
Know which ride to take, what traffic looks like now, and when a different boarding point or departure time might help.
Unsa nga jeep ang sakyan padung Divisoria karon?
Safe by design
The AI reads the road. An engineer makes the call.
Gemini reads the scene and explains each recommendation. A separate, predictable optimizer sets the timings. A language model never drives the signal.
- Minimum green time on every approach
- Fixed yellow and all-red intervals
- Enough time for pedestrians to cross
- A maximum wait, so no lane gets starved
RolloutOne step at a time
AdvisoryPilot
SimulatedDemo
LiveLater
Live control only comes with city authorization and hardware integration. Until then, nothing touches a real signal.
Everything in one place
Nothing bolted on. One live picture for everyone.
Vehicle density
Lane-by-lane counts from video, with jeepneys, buses, cars, motorcycles and motorelas counted separately.
Signal controller
Recommends timings that clear the heaviest lanes first, and falls back to the fixed plan if a camera fails.
Command dashboard
A map colored by density, with wait times, flood alerts, sandbox-tested emergency overrides and an approval log.
City sandbox
Test plans on a model of Divisoria against rush hour, a flood closure, an event or an ambulance.
Commuter guidance
Later update
Which ride to take, what traffic looks like now, and a better boarding point or time to leave.
Traffic assistant
Ask about traffic in Bisaya, Tagalog or English, whether you work for the city or ride the jeep.
Common questions
Before you ask, you’re probably wondering…
Does it control traffic lights today?
No. In the pilot it recommends and a city engineer decides. Approved plans run in a simulation. Connecting to real signals is a later phase that needs city authorization and hardware integration.
Will it give drivers tickets?
No. Ticketing and enforcement are out of scope for the pilot.
Does it identify people or plates?
No. There’s no license plate reading and no face identification. Faces and plates are blurred, we keep no raw footage beyond what the pilot needs, and we follow the Data Privacy Act. We get permission before using any camera feed.
What happens if a camera fails?
If a feed fails or the AI isn’t confident, the system falls back to the existing fixed-time plan.
How accurate are the counts?
Counts are shown as ranges and are measured against manual counts from the same footage.
Will it reduce traffic across the city?
We’re not claiming that. Comparisons of fixed timers and adaptive plans here are simulated, and a measured reduction is the question a real pilot would answer.
Is there an app for commuters?
Not yet. We’re building the traffic system first, and ride and route guidance comes in a later update. Whether its suggestions ease traffic is a hypothesis we’ll test, not a proven effect.
Divisoria first. Then the rest of the city.
We’re proposing a pilot at two to four Divisoria intersections, still to be confirmed. No citywide control, no live hardware and no ticketing - just better evidence for the people who time the signals.