Speed

Of everything a driver monitoring system measures, speed has the clearest relationship to both collision likelihood and severity.

Why it is the strongest measure

Severity scales steeply with impact speed, which is physics rather than statistics.

Stopping distance grows faster than speed does.

And the association with collision involvement is consistent across studies in a way that harsh braking is not.

Which makes speed the one measure worth building a programme around, if only one is chosen.

Measuring it badly

Speed derived from successive positions carries error that grows with the sampling interval and is worst exactly where it matters — during rapid changes.

Speed read from the vehicle bus is a measurement.

Ask which yours does, because a speeding event derived from position is contestable and one read from the vehicle is not.

The speed limit problem

A speeding event requires knowing the limit, which comes from a map database.

Map limits are wrong more often than people assume: recent changes, temporary limits, roadworks, and roads where the database has no entry.

Which produces events on roads where the driver was compliant, and those events destroy confidence in the whole system.

Check a sample against reality before using speeding data for anything, and expect to find errors.

Speed against conditions

The limit is a maximum, not a target.

Driving at the limit in fog, rain or heavy traffic is unsafe and is not detected by any system that compares against a map figure.

Which means a clean speeding record does not mean appropriate speed, and a programme that reports only limit exceedances is measuring the easy half.

The schedule connection

A route that cannot be completed in the time allowed produces speeding, reliably.

Which means a speeding pattern concentrated on one round is a scheduling finding, and coaching the drivers on it is treating a symptom.

Check the achievable time against the allowed time before the conversation, because if the round is impossible everyone driving it will speed.

What to do with it

Report exceedance severity and duration, not just count — ten miles an hour over for two minutes is different from two over for an hour.

Investigate patterns by route before by driver.

And use it for coaching rather than for ranking, because the ranking is dominated by which roads someone drives.

Check a sample against real limits

Before using speeding data for anything individual.

Fifty events, verified against the actual limit at that location.

Map databases are wrong on recent changes, temporary limits and minor roads.

An error rate above a few percent means the data cannot support a conversation about a person, and the affected roads should be excluded until corrected.

Connect policy and data

The choices in this note can be compared with the agency workflow page. Keep the written purpose in control and enable only the information needed at this boundary.

Independent reference

For a thematic point of reference, see the International Organization for Standardization. This established source is a useful independent reference for the issue discussed here.