The encyclopedia · Product & Design · Product decision · 2007–2012
TomTom turned every GPS car into a traffic sensor, ending the need for road detectors
Floating car data let TomTom map traffic from drivers themselves—no new roadside hardware anywhere.
TomTom
the move
Traditional traffic services depended on roadside detectors or cameras—expensive to install and only where agencies could afford them. TomTom's floating car data (FCD) approach, introduced with HD Traffic in 2007, treats each connected vehicle as a moving probe.
In a 2012 paper, TomTom's Nick Cohn and Hugo Bischoff describe the pipeline: raw GPS positions with timestamps are map-matched onto the road network, then aggregated with data from mobile phones and Bluetooth. Their case-study dataset alone contained about 150 million measurements; accuracy depends on sample size and map quality.
Independent research by the Dutch government's NDW compared TomTom's floating-car travel times with fixed ANPR cameras and reference systems, concluding the FCD was accurate and highly useful for traffic management—while exposing how data from the wrong fleets or too-small samples can mislead.
why it works
- Probe coverage scales with users, not construction budgets
- Map-matching turns noisy GPS into road-level speeds
- Fusing phone signals fills gaps where GPS is sparse
- Benchmarking against fixed sensors proved the data quality
what transfers
If you cannot afford to instrument the world, instrument the users: data harvested from customers' existing devices can substitute for purpose-built infrastructure.
what came after
Floating car data became the industry standard for real-time traffic: TomTom's HD Traffic expanded across Europe and the US, and governments began buying probe data instead of building detectors. The same sample-based logic now underpins most navigation traffic services.
references
- Floating car data for transportation planning (NATMEC 2012)
- The Quality of Floating Car Data Benchmarked: An Alternative to Roadside Equipment?
- Mobilnavis nutzen zunehmend Nutzerdaten (HD Traffic, mobile-phone data)
spotted an error? The archive wants to know.