Forklifts and pedestrians: measuring co-activity by zone, without watching people
About 8,300 lost-time accidents a year involving forklifts, around ten deaths, half by lateral tip-over. What cameras can count, and what they must not do.
The INRS counts about 8,300 lost-time occupational accidents a year involving powered industrial trucks, 580 permanent disabilities and around ten deaths, half of them by lateral tip-over.1 A more recent note from the same institute puts the average at 13 deaths a year, two thirds by lateral tip-over.1 Behind these national figures, the mechanism is always local: an aisle, a crossing, a speed, a raised load, a pedestrian where they should not be.
These situations are filmed in almost every warehouse. The question is what to do with that, and above all what to refuse to do with it.
Two cases, two lessons
In Saint-Priest, near Lyon, a 25-year-old temporary worker was found trapped under his overturned forklift, with no witness, and had a foot amputated; the year of the accident remains to be confirmed in our sources.2 The detail that matters is not the tip-over, it is “no witness”: nobody was in the zone, nobody saw, nobody could raise the alarm.
In Pamiers, in July 2026, a forklift operator died while moving tanks when his truck overturned; the prosecutor and the labour inspectorate are investigating.2 It is a factory, not a warehouse, but the ingredients are the INRS ones: the height of the load, the turn, the floor.
In both cases, what a camera would have shown is not a person’s fault. It is the organisation around them: lone work in a zone, co-activity at a given hour, traffic outside the marked lanes.
What can be measured without monitoring anyone
The reflex of “intelligent” video is to follow individuals: who drives too fast, who cuts across the marking. That is exactly what we refuse, because it is ineffective for prevention and unacceptable for a works council. What prevents an accident is a number per zone, not per person.
| What the site writes | What the agent counts, per zone and time slot | What it changes |
|---|---|---|
| “Forklift and pedestrian at the same time in aisle 4” | Number of co-presence episodes, their duration, their hours | A flow is moved, a picking round is shifted |
| “Lone work in the high-bay storage zone” | A single person present for more than ten minutes | Lone-work organisation is reviewed |
| “Traffic outside the marked lanes” | Episodes of forklifts off the marking, per zone | A worn marking or a shortcut turned habit is fixed |
| “Pedestrian in the forklift lane” | Episodes of pedestrian presence in the lane, per zone | A protected walkway closed “temporarily” is reopened |
Every line goes into the risk assessment document (DUERP), never into an individual file. Stored images are blurred, no person is identified, and questions about an employee are refused by design. That is what makes the tool presentable to a works council, and that is what makes it useful: you correct a zone, not an employee.
Coverage, the indicator everyone forgets
The Saint-Priest case raises a question few sites can answer: which zones are actually seen, and at what hours? Camly measures, for each monitored zone, the share of working time during which the check took place. A zone with low coverage is not a safe zone, it is a blind one. That single number often justifies moving a camera, or adding one, before any other spend.
What Camly does not do
- Camly does not stop a forklift and does not prevent a tip-over to the second. Equipment, speed limiting and operator training do.
- A periodic check catches what lasts and what repeats. It does not replace an anti-collision device.
- No indicator is produced per person. No facial recognition, no individual tracking, no join with badges or the warehouse management system.
Where to start
Take your site’s traffic plan and mark the three crossings where the teams tell you there is “always someone”. Check that the cameras see them. Write for each one the sentence that describes the co-activity to avoid. Thirty days later you will have hours, durations and recurrences per zone: that is the scope of a Camly pilot, and the first time the DUERP will talk about forklifts with the site’s own numbers.
Sources
- INRS, powered industrial trucks (ED 125): about 8,300 lost-time accidents, 580 permanent disabilities, around ten deaths a year; INRS technical note NT 46: 13 deaths a year on average, two thirds by lateral tip-over. ED 125 (PDF), NT 46 (PDF)
- Lyon Capitale, forklift overturned on a logistics site in Saint-Priest, foot amputated (year to be confirmed); info.fr, forklift operator killed in Pamiers, 3 July 2026. lyoncapitale.fr, info.fr