As a strategic partner working across the Google Maps Platform ecosystem, the question we hear most often isn’t whether mobility data exists — it’s how to actually use it. Google Roads Management Insights (RMI) offers one of the clearest answers to that question.
In this blog, we examine Google’s data ingestion (BigQuery/Pub/Sub), traffic prediction methods (graph neural networks, map-matching, change detection), and privacy controls.
Google Roads Management Insights (RMI) is used by public agencies and private organizations that depend on road network performance. This includes departments of transportation (DOTs), municipalities, regional planning organizations, logistics providers, utilities, and infrastructure operators. These organizations use RMI as a scalable mobility data layer that integrates into existing systems and supports consistent measurement across their networks.
Example: Managing Seasonal Traffic in Mountain Corridors
Google highlights an example where the Colorado Department of Transportation uses RMI to analyze traffic patterns along the I-70 mountain corridor, a critical and often congested route. By identifying when and where bottlenecks occur, CDOT can better plan for seasonal demand and allocate resources proactively, improving reliability along one of the state’s most important transportation corridors.
This shift from reactive traffic management to trend-based planning is exactly where RMI delivers value, turning historical mobility data into actionable insight at network scale.
At a technical level, RMI delivers two core outputs:
- Historical travel time patterns across defined routes
- Near real-time speed interval updates based on live conditions
These are derived from aggregated, anonymized activity across Google Maps and Waze, continuously processed through Google’s cloud infrastructure.
The Google Maps Platform’s RMI is a geospatial analytics service tailored to road authorities. You define the routes you care about (e.g. major corridors, detour paths) using the Roads Selection API. Google then continuously gathers anonymized probe data from Maps and Waze users to calculate two things: the periodic average travel time, and the instantaneous “speed reading intervals” along each route. These data are delivered to you via Google Cloud: the historical time series lands in a BigQuery dataset (shared via Analytics Hub) and live updates stream through Pub/Sub. With just a few API calls, a DOT can obtain continual travel-time feeds without deploying new sensors.
How RMI relates to Roads API, Routes API, Waze for cities, Street View Insights, and Earth Engine
Roads Management Insights (RMI) offers aggregated and real-time traffic analytics for user-defined routes. The Roads Selection API lets agencies “subscribe” to routes; Google then periodically publishes travel durations and segment speed intervals to BigQuery or via Pub/Sub. The Roads API (“Snap to Roads,” speed limits) and Routes API (ETA and directions) supply foundational data. Waze for Cities enables sharing of crowdsourced incident data (crashes, hazards) via free GeoRSS/XML feeds updated ~2 minutes. Street View Insights (in Vertex AI) exposes Street View imagery attributes (utility poles, signs) for asset analytics. Google’s satellite/Street View imagery and Earth Engine datasets (now in BigQuery) add raster context. Integration via BigQuery Data Exchange (Analytics Hub) and Cloud Pub/Sub links live/archival feeds into DOT analytics pipelines. All services feature Machine Learning enhancements (e.g. graph neural network (GNN) traffic forecasting, congestion estimation, and Vertex AI image analysis).
Data Flow Architecture: The system architecture shows Google’s managed data flows feed a connected ecosystem
| Google Service | Data Provided | Typical Use |
|---|---|---|
| Roads Management Insights (RMI) | Periodic & real-time travel time, speed intervals for custom routes | Traffic trend analysis, incident response, planning |
| Roads API (Maps Platform) | Nearest roads, snapped coordinates, posted speed limits | GPS trace map-matching, enforcement, planning |
| Routes API (Maps Platform) | Route geometry, driving ETA (with traffic) | Performance metrics (DOT indicators), congestion management |
| Waze for Cities | Real-time incident/hazard feeds (JSON/GeoRSS) | Incident detection, event response, traveler info |
| Street View Insights | Asset presence/attributes (poles, signs) via imagery analysis | Asset inventory (pole location/conditions), compliance audits |
| Earth Engine datasets (now in BigQuery) | Petabyte satellite imagery and land data (e.g. elevation, land cover) | Vegetation analysis, environmental risk, change detection |
Incident Detection and Emergency Response: Ingest Waze and RMI real-time alerts to flag accidents or congestion; map integration with ATMS systems enables dispatch optimization. For example, by streaming RMI data into an ATMS, traffic managers can track clearance times and compare performance against FHWA TIM (Incident Management) benchmarks. In planning, historical RMI trends help prioritize projects: if a segment’s travel time index is consistently high, it might need widening or signal upgrades.
Planning and Maintenance Prioritization: RMI’s historical travel-time trends help identify chronic bottlenecks or safety hotspots.
| Use Case | Metric/KPI | Google Data & Tools |
|---|---|---|
| Incident response | Incident detection time, clearance time | Waze alerts; real-time RMI Pub/Sub; integrate into TMC dashboards. |
| Congestion analysis | Travel time indices, reliability (e.g. Buffer Index) | RMI historical vs. real-time travel durations. |
| Maintenance Planning | Frequent impairment locations, pavement condition ratings | Street View/imagery analysis for pavement defects; Earth Engine layer changes. |
| Asset inventory | Asset count/density (poles, signs) | Street View Insights (Vertex AI) with approved models. |
| Emergency evacuation | Route clearance, evac time savings | Real-time RMI to simulate detour efficiency, up-to-date speeds. |
BigQuery & Pub/Sub Costs:
Google Maps Platform is pay-as-you-go by billable call/event (Routes, Roads). RMI historically was a pilot (special pricing) but now needs a contract. BigQuery charges by bytes processed (first 1 TiB free), or flat-slot. Pub/Sub charges ~$40/TB after first 10 GB. Waze for Cities is generally free for public agencies (data-sharing). Note: exact enterprise SLAs and pricing tiers are unspecified, contact Sanborn Spatial Intelligence for a certified quote.
| Google Service | Data Provided | Typical Use | Access/Notes |
|---|---|---|---|
| BigQuery & Pub/Sub | Scalable data warehouse; streaming ingestion | Data fusion, analytics pipelines, custom KPIs | BigQuery pricing: on-demand or flat-rate slots (first 1 TiB free). Pub/Sub: pay per TB/messages (first 10 GB free). Real-time RMI uses Pub/Sub to push updates. |
What about AI?
AI is central to how Google Roads Management Insights delivers value. Traffic predictions are powered by graph neural networks (GNNs) that combine real-time signals with historical patterns to generate accurate travel time and congestion estimates. These models are built using TensorFlow and informed by research from Google DeepMind.
At the same time, Vertex AI enables automated feature extraction from imagery, supporting use cases like asset identification and condition monitoring from Street View data.
Take the Next Steps: A Conclusion to RMI
As a Premier Google Partner, Sanborn Spatial Intelligence helps define our clients’ system architecture (“the integration handshake”), ensuring compliance with Google’s geospatial terms and optimizing API usage to minimize cost within all of Google Maps Platforms APIs and SDKs. We also monitor Google’s Usage Reports to track calls and billing, avoiding surprises.
Together, we help our clients turn raw data into actionable intelligence: faster response to incidents, smarter maintenance, and more efficient planning. By following best practices, monitoring costs, and staying compliant with terms and laws, you can leverage Google’s global data at scale.
Key Interoperability Notes: Google’s terms strictly forbid data scraping, caching, or mixing with non-Google maps. For example, DOTs must not “trace or digitize” Google content (e.g. building polygons from Satellite, or rehosting Street View imagery). However, RMI analytics content (durations, intervals) can be exported/cached within limits (contact Sanborn’s Spatial Intelligence team for more information).
