80% fewer warehouse floor incidents for a global logistics operator
A forklift and a person meeting in the same aisle is the collision warehouse safety teams fear most, and the hardest one to see coming. This is what happened when a global logistics operator pointed Invigilo at the dock and aisle cameras it already owned.
The results, up front
Two outcomes measured on the operator's warehouse floor, shown next to a platform-wide accuracy figure so the numbers carry their own context.
Forklifts and people share the same floor
None of what follows is unique to this operator. It is the operating reality of warehousing everywhere. Trucks arrive on a schedule the floor does not control, pick rates climb toward the peak, and forklifts and pedestrians end up sharing the same aisles, the same dock doors and the same blind corners for entire shifts.
Most of the danger never makes it into a report. A driver brakes hard, a picker steps back, both carry on with the shift. Near misses like these happen daily in a busy facility and are almost never written up, so the safety team ends up planning around the small fraction of events it actually hears about.
Supervision cannot fix this on its own. No supervisor can watch every dock and corner at once, and when an incident does happen, the investigation usually leans on memory and secondhand accounts rather than footage of what occurred. That was the starting point here, as it is in most of the sector.
SafeKey on the cameras already there
The operator did not install new cameras. SafeKey, the AI video analytics module of Invigilo SafeSuite, connected to the existing dock and aisle CCTV. Any camera with an RTSP stream can feed it, and a standard site is live within 24 hours of camera access.
Each area got its own rules. Speed zones on the transit routes, pedestrian exclusion zones at dock doors and staging lanes, and vehicle-person proximity rules wherever the two have to mix. The detections were drawn from a library of 40+ risk types built on 60+ computer vision models, tuned to how each part of the floor actually works.
When a rule is broken, the alert reaches the shift manager's phone through WhatsApp, Microsoft Teams, Telegram or email in under a second, with escalation chains behind it if nobody responds. Every detection is kept as a timestamped clip, so the event exists as evidence rather than as an anecdote.
Existing CCTV, no new hardware
Dock and aisle cameras connected over RTSP. A standard site is live within 24 hours of camera access.
Zone rules per area
Speed zones, pedestrian exclusion zones and vehicle-person proximity, each set for the way that part of the floor is used.
Alerts on the floor
Sub-second notifications to WhatsApp, Microsoft Teams, Telegram or email, with escalation chains when an alert goes unacknowledged.
Evidence by default
Every detection becomes a timestamped clip that feeds toolbox talks and corrective actions.
What changed on the floor
The two measured results are easy to state and worth unpacking. Incident response got 30% faster for a simple reason: the alert reaches the floor while the situation is still live. A shift manager who hears about an overspeed event in under a second can walk over and deal with it now, instead of reconstructing it at the end of the week.
The 80% reduction on watched warehouse floor risks took longer, and it came from two directions. Drivers and pickers adjust their behaviour once they know the aisles are watched and every event produces a clip. And the clips themselves showed where the floor layout was the real problem, so recurring hotspots got engineering fixes: rerouted walkways, relocated staging, firmer separation at the dock doors.
The biggest shift is harder to put a percentage on. Before the deployment, near misses on this floor mostly went unrecorded. After it, they were counted, timestamped and patterned by aisle and by shift. That visibility is uncomfortable at first, because the numbers start higher than anyone expected. It is also the entire point: a risk that is never seen never gets fixed.
The loop actually closes
Detection on its own changes nothing. Each step from camera to correction has to hand off to the next, and this deployment worked because every link in that chain is part of the platform.
It watched continuously
Detection runs around the clock on every connected dock and aisle camera, at 85%+ verified accuracy, without fatigue and without looking away.
It told someone in time
Sub-second alerts with escalation chains mean the right person hears about the event while it can still be interrupted.
It kept the evidence
Timestamped clips replaced memory in investigations and gave toolbox talks something concrete to show the next shift.
It respected the workforce
No facial recognition. Long-standing ISO 27001 certification with GDPR and PDPA compliance, deployed edge, on-premise or hybrid so footage stays under the operator's control.
See it on your dock cameras
A proof of concept on one dock or aisle, live within 24 hours of camera access.