Every transport operator can produce a report showing that duty hours were within limits. Almost none can produce one showing whether the person driving was actually fit to do it.
Those are different questions, and only one of them is answered by the tachograph.
What hours rules actually control
Drivers’ hours legislation is a ceiling on exposure. It bounds daily driving time, continuous driving, and minimum rest. That is genuinely useful — an unbounded system produces exhausted people quickly.
But a ceiling is a blunt instrument. Within the same compliant limits you can have:
- A driver on a settled, consistent day shift.
- A driver who has just rotated from nights onto earlies with the minimum rest in between.
- A driver in their fourth consecutive early start, each beginning before five.
- A driver whose rest period was legally sufficient and practically useless because it was taken in daylight after a night run.
All four are compliant. They are not equivalently safe, and every operator knows it.
The factors the limits do not see
Fatigue research is consistent about what drives risk, and most of it sits outside what hours rules count:
- Time of day. The early hours are the worst window regardless of how much sleep preceded them. Two till six is not the same as two till six in the afternoon.
- Rotation direction and speed. Moving forwards through shifts is easier to adapt to than backwards, and rapid changes are worse than either.
- Cumulative load. The fifth long shift in a row is not the same as the first, even where each is individually within limits.
- Quality of recovery, not just its duration. Rest taken at the wrong time of day does less.
- Commute. An hour each way is not duty time and is very much awake time.
None of that appears in a compliance report, which is why a compliance report is a poor proxy for risk.
Self-declaration is not the fix
The common response is to ask. A form at the start of a shift: are you fit to work?
It fails for reasons that have nothing to do with honesty. People are poor judges of their own impairment when fatigued — that is a symptom, not a character flaw. And the person answering knows what happens to the shift, and to their pay, if they say no. A question with a costly answer gets the cheap one.
Which does not mean discarding the driver’s own view. It means not resting the whole control on it.
Scoring the shift instead
The alternative is to assess every shift from the roster you already hold, before it is worked. The pattern is known in advance: start time, length, what preceded it, how much recovery there was, how many consecutive shifts, how the rotation is moving. That is enough to produce a risk score without asking anyone anything.
The value is that it changes when the conversation happens. A high score on Wednesday’s roster can be fixed on Monday by moving one duty. The same fatigue discovered on Wednesday morning is an argument at a depot gate with nobody to cover.
We built RailGard AI on exactly this basis for rail: every shift scored against time of day, shift length, weekly hours, recovery time and night working, with each contributing factor logged so a score can be defended rather than taken on trust. The regulator differs across transport modes; the physiology does not.
The uncomfortable part
Doing this properly means occasionally generating evidence that a legal roster is a risky one. Some operators would rather not know, on the reasoning that what is not recorded cannot be produced later.
That calculation has never held up well. The pattern that produced an incident is reconstructible afterwards from the rosters and the tachograph data you were already required to keep. The choice is not whether the evidence exists. It is whether you saw it in time to act.
Rostering against fatigue rather than around it? Get in touch, or read how we build for travel and transport.