RMI's biggest release since launch.
Three new feeds, one bigger product.
In the first week of May 2026, Google extended Roads Management Insights with three new experimental feeds: vehicle counts, harsh-braking events, and user-reported incidents. Each has been quietly requested by every public-sector customer we've spoken to since RMI launched. Stacked together, they take RMI from a great speed-and-travel-time feed into something much closer to a full operational picture of a road network, and they unlock a new generation of safety, capacity, operations, and simulation use cases that were previously out of reach.
What's in each feed.
Vehicle counts, hourly, per road segment. RMI now tells you how many vehicles passed each stretch of road in the last hour, anonymised and aggregated. The dataset that finally answers "how much traffic does this corridor actually carry?" without a roadside sensor or a manual count.
Harsh-braking events, anywhere on the road network. RMI now surfaces anonymised harsh-braking events as point data, wherever drivers are slamming on the brakes, whether that's at an intersection, on a sharp curve, near a lane drop, or anywhere in between. The most direct safety signal we've had at network scale, visible across every road in your selection.
User-reported incidents: accidents, hazards, stopped vehicles. The same incident reports Google Maps users send from the app, accidents, hazards, slowdowns, stopped vehicles, now available as a real-time feed for your network. Because there are millions of Google Maps users on the road every day, you catch incidents your cameras and helplines never will.
What this release unlocks.
These feeds are useful on their own. Each one opens a set of use cases that used to need expensive hardware, slow surveys, or a separate vendor. Here's what we've either deployed or scoped with a customer since RMI launched.
Vehicle counts
- Capacity and corridor sizing, know exactly how much traffic each road carries, hour by hour, with no roadside hardware
- Signal-cycle tuning, retime signals for peak vs off-peak using real flow numbers, not survey estimates
- Before-and-after studies, measure the impact of any change, a new signal, a lane reversal, in days not quarters
- Development-impact studies, see how much new traffic a mall, office, or housing project adds to surrounding roads
- Toll and revenue forecasting, the hourly count data toll authorities have always asked for
Harsh-braking events
- Proactive safety fixes, find risky spots, intersections, curves, lane drops, before the crashes show up in police data
- School-zone and pedestrian audits, a city-wide safety review, not just where sensors happen to be installed
- Insurance-claim verification, cross-check claims against the actual braking pattern at the reported location
- Speed-limit review, a data-driven case for where speed limits genuinely need to change
User-reported incidents
- Real-time alerts to operations, shrink the citizen-call to dispatch loop from over an hour to under five minutes
- One feed for all reports, replace the patchwork of helpline, social media, and WhatsApp inbound with a single source
- Smarter emergency routing, see corridor status and existing reports before the response unit arrives
- Automated travel advisories, push verified incident data to nav apps, dashboards, and public alerts from one source
Now you can model the future, not just measure the present.
Until this release, RMI told you what is happening on your roads and what usually happens. With vehicle counts and incident feeds added, the same pipeline now powers genuine forecasting and what-if models, no separate sensor procurement, no commissioned simulation study.
- What-if modelling , close a corridor for repairs, simulate the impact on the surrounding network
- Special-event planning , concerts, marathons, processions, religious gatherings with real capacity numbers
- Pre/post project evaluation , fund what works, retire what doesn't, with evidence the audit office can defend
- Climate-event preparation , monsoon, snow, and flood contingency plans grounded in real flow data
Notes from our deployments.
Vehicle count cadence is hourly. Well-suited for capacity studies, signal-cycle tuning, and before-and-after analysis. Not real-time, if your use case needs minute-level vehicle flow (e.g. adaptive signal control), you still need sensors.
Harsh-braking events arrive as anonymised clusters, not individual driver events. This is a privacy feature, not a limitation, the cluster signal is exactly what you want for road-design risk analysis.
The signal volume from Google is high, we filter it down to what matters. The raw feeds are dense. We've built an alerting algorithm that surfaces only the events worth acting on, so your operators see the signal and not the noise. The same filtering can be tuned per city, per corridor, and per operating team.
For the long-form treatment of how each feed integrates into TraffiCure deployments, see the "Brand new in May 2026" section of the full RMI guide.