How it works · The mechanism
From Google probe data to a live command centre
Trafficure is the activation layer on Google Maps Roads Management Insights. Aggregated anonymized movement from 1B+ phones becomes a defined road network, scored against its own history every two minutes, and delivered as alerts, dashboards and reports. No cameras, no sensors, no construction.
Get a demo →The short version
RMI is the plumbing. Trafficure is the activation layer.
Google Maps Roads Management Insights delivers anonymized road-network traffic data to public-sector operators. On its own that data arrives as BigQuery tables and Pub/Sub feeds.
Trafficure turns those raw feeds into the part a city actually uses: a defined network, baselines, live congestion alerts, dashboards and automated reports. Lepton is the authorized RMI reseller in India and the UAE, so the data layer and the activation layer come from one place.
Stage 1 and 2, the input
Where the data comes from, and how the network is drawn
- 01
Source · Aggregated anonymized movement
Speed and movement signals from 1B+ Google Maps and Waze devices, aggregated per road segment. Continuously available since 2009, with no hardware to install on the ground.
- 02
Define · The Roads Selection API
An agency sets exactly which road network it monitors through the Roads Selection API. That selection is the network Trafficure watches, from a few corridors to a whole city.
- 03
Privacy · Privacy by design
No personal information, ever. The data is aggregated with k-anonymity applied at source by Google, designed to align with GDPR and India's DPDP Act.
Stage 3, the feeds
Two streams: history and the live present
Once the network is defined, two feeds keep it current. Historical travel-time patterns arrive via BigQuery and give each road its normal profile. Near-real-time speed intervals arrive via Pub/Sub and refresh the whole network roughly every two minutes.
Together they let Trafficure say not just how fast a road is moving, but whether that is normal for this road at this time.
Probe to alert
Probe to alert in two minutes.
One continuous route: phones to a defined network, history and live speed converging on the analytics engine, the full cycle under 120 seconds, then alerts, dashboards and reports. No part of it waits on a camera.
Stage 4, the analytics
Normalized against the road graph, scored in under two minutes
Live speeds are normalized against the road graph, 15,000+ segments for a city, and scored against each road's ~90-day baseline. When a segment drifts far enough from its normal profile, that is a congestion build-up, and it is surfaced. The full analytics cycle runs in under 120 seconds, which is why alerts arrive while a jam is forming rather than after it has set.
- 01
Normalize · 15,000+ segments per city
Raw probe speeds are mapped onto the road graph segment by segment, so every monitored road has a current, comparable reading.
- 02
Score · Deviation from baseline
Each segment is compared to its own ~90-day normal. Alerts fire on the deviation, not on a fixed speed threshold, so a slow arterial and a fast bypass are each judged on their own terms.
- 03
Health · CityPulse scores
Segment readings roll up into composite traffic-health scores that feed the dashboards and the automated reports.
KPI dashboard
Corridor speed, congestion trend and live alerts
The Trafficure command centre renders the scored network as a live KPI dashboard. Speed tiles show the current congestion ratio per corridor against its baseline. A trend chart traces the CityPulse score over the monitoring window.
An alert feed lists active incidents by severity with the delay versus free-flow speed for each segment, so an operations team can prioritize without reading through raw data.
Stage 5, the delivery
Where the answer lands
The scored network is delivered through the channels an operations room already lives in. The command-centre dashboard runs in the browser with nothing to install.
Alerts route to WhatsApp, SMS, email, Slack, a webhook or X, with configurable thresholds, auto-resolve and priority levels. Reports generate as PDFs, and data exports in standard formats with no vendor lock-in.
What changes
What the two-minute cycle changes
- 01
Alerts before the jam sets
A refresh every two minutes scored against ~90-day baselines means a build-up is flagged as it forms, not learned from a citizen complaint 15 to 90 minutes later.
- 02
The whole network, not the cameras
Because the signal is probe data, every road an agency selects is covered, not only the junctions a camera happens to face.
- 03
Live in weeks, not tenders
With no hardware to procure or build, a city goes live in 2 to 4 weeks, or 4 to 8 weeks from signature when starting a fresh RMI subscription.
The honest limits
What this mechanism is not for
The same design that makes Trafficure software-only sets its edges, and we name them plainly. Probe data is excellent for network monitoring and analysis, and it is not a sub-second feed, so it is not the layer that drives the traffic lights at a junction.
There, sensors win, and camera ITMS stays the right tool for enforcement and number-plate recognition. And because k-anonymity publishes a road only when enough vehicles are present, very quiet roads can drop out of coverage. Trafficure monitors the network ITMS cannot see, alongside the junctions ITMS already runs.
- 01
Not for · Driving the lights at a junction
Timing traffic lights second by second needs sensor-grade latency. Trafficure is the network-monitoring layer, complementary to the camera and sensor systems at the junction.
Proof
Running this way in a live city
The pipeline above is what runs in Pune, the first Indian city with software-only city-wide traffic management, deployed with Pune City Traffic Police and Google India. The same mechanism is live across the Kolkata region.
- Pune City Traffic Police
- Pune Municipal Corporation
- Kolkata Police
- Howrah Police
- Bidhannagar
- Barrackpore
See the pipeline run on your own city.
We define your network with the Roads Selection API and demo Trafficure on your actual roads, not slides.
Get a demo →