In short: Around 1,300 UK workers a year are seriously injured in forklift accidents, and most of them are pedestrians. A new generation of forklift safety systems, proximity warning, AI pedestrian detection, geofenced speed control, can cut that number, but each one depends on radio coverage across the whole site, including the yard, the racking aisles and the trailer, which is exactly where factory WiFi gives up.
Key Takeaways
- Forklifts injure about 1,300 UK workers a year — UK Material Handling Association figures put fork lift trucks in roughly a quarter of workplace transport injuries, and around 57 per cent of those hurt are pedestrians rather than drivers.
- The useful safety tech is a live position layer — proximity warning, geofenced speed zones and near-miss logging all depend on knowing where every truck and every person is, continuously and site-wide.
- Moving vehicles break factory WiFi — access-point handover drops connections at exactly the moment a truck is moving fastest, which is why forklift safety systems demo well and roll out badly.
In a nutshell
How dangerous are forklifts in UK factories?
More dangerous than almost anything else on the floor. The UK Material Handling Association's analysis of RIDDOR data puts fork lift trucks in about a quarter of all workplace transport injuries, with roughly 1,300 workers a year suffering serious injuries, and the Health and Safety Executive's fatality statistics show being struck by a moving vehicle killing around 25 workers a year across British workplaces. The detail that shapes everything else is who gets hurt: in the UKMHA's figures, around 57 per cent of forklift injury victims are pedestrians, colleagues walking the same aisles the trucks drive.
The injuries are also disproportionately severe. A counterbalance truck weighs three to five tonnes unladen, carries its load high, and reverses for a large share of its working cycle with the driver's view partly blocked by the mast. A collision at 10 km/h that would dent a car crushes a foot. HSE prosecutions after forklift strikes routinely end in six-figure fines, and the pattern in the prosecution reports is monotonous: a pedestrian in a vehicle area, a driver who could not see them, and a segregation plan that existed on paper.
That last point matters for what follows. The HSE's guidance, HSG136 on workplace transport safety, is clear that the hierarchy starts with keeping people and trucks apart: barriers, marked walkways, separate doorways, one-way systems. No technology substitutes for that, and a plant that has not done the segregation work should spend its budget there first. The technology case begins where segregation genuinely cannot be complete, which in a working factory with shared aisles, cross-docking and mixed traffic is most of the floor.
What counts as a forklift safety system in 2026?
The market sorts into four tiers by how much infrastructure they need.
The first tier needs none: blue spot and red zone projection lights, audible reversing alarms, mirrors and camera-monitor kits on the truck itself. These are cheap, effective within their limits, and every fleet should have them. Their limit is that they inform whoever happens to be looking.
The second tier is truck telematics: access control so only trained drivers can start a truck, pre-shift check prompts, impact detection that logs the collision the shift never reported. Fleet providers fit this as standard now, and the data alone changes behaviour; drivers who know impacts are logged have fewer of them.
The third tier is active proximity warning. Pedestrians carry a tag, trucks carry a detector, and both parties get warned, by vibration, light and sound, when the distance closes, typically using ultra-wideband radio, which ranges reliably to about 30 centimetres through racking and around blind corners where cameras and eyes fail. Systems in this class, Linde's Safety Guard among the best known, can also slow the truck automatically inside a defined bubble.
The fourth tier is AI pedestrian detection and real-time location. Camera systems such as Blaxtair distinguish a person from a pallet and warn only when it matters, which preserves the alarm's credibility. Site-wide real-time location systems (RTLS) go further: every truck and tagged person on a live map, geofenced speed limits that slow trucks automatically near doorways and packing benches, and a recorded stream of near misses, the encounters that HSG136 asks sites to learn from and that almost no site currently captures.
Why do forklift safety pilots stall at rollout?
The first three tiers mostly live on the truck and work anywhere. The fourth tier is a networked system, and this is where a familiar pattern appears, the same one we have described in condition monitoring: a successful trial in one aisle, followed by a rollout that dies in procurement because the plant's wireless cannot carry it.
The technical reasons are specific to moving vehicles. A forklift crosses a WiFi access point's cell in seconds, and the handover between access points, tolerable for a laptop, produces multi-second dropouts on a fast-moving client. A safety function that slows a truck near pedestrians cannot tolerate a dropout at 15 km/h; the vendors know this, so they either fit their own dedicated radio mesh, another network for the plant to own, or quietly downgrade the feature set. Steel racking, which fills the building, creates the same reflective, shadowed radio environment it creates for everything else. And the yard, where trucks meet lorries and pedestrians meet both, usually has no coverage at all, despite being where the HSE's fatality statistics concentrate.
The honest conclusion is that tier-four systems are network projects wearing safety branding, and should be procured as both at once. A site that evaluates the radio layer with the same rigour as the safety layer avoids the stalled pilot entirely.
What does a live position layer make possible?
Once position data flows continuously, the applications compound. Geofenced speed control is the headline: trucks capped at walking pace within ten metres of a pedestrian door, automatically, with no reliance on the driver remembering. Dynamic exclusion holds trucks out of an aisle while an order picker works at height in it. Near-miss analytics turn the safety committee's anecdotes into a ranked list of the five worst interaction points on site, with timestamps, which is the evidence that justifies moving a walkway or re-timing a shift overlap.
The same layer pays for itself outside the safety case, which is often what gets the capital signed off. Live truck location feeds utilisation numbers, and most sites that measure discover they own two more trucks than they need at a five-figure annual rental each. Impact data attributed to a truck, a place and a time cuts damage bills and insurance premiums. The safety case and the operations case are the same infrastructure viewed from two budgets.
Wi-Fi or private 5G for a moving fleet?
Both can work, and the right answer depends on the site, so it is worth being precise about the trade-off rather than partisan. A modern Wi-Fi 6 network, professionally surveyed, with fast-roaming configured and dense access-point placement, can support RTLS and telematics indoors, and if a plant already has one, the marginal cost argument favours using it.
Private 5G earns its place on three conditions that manufacturing sites meet more often than they expect. Where vehicles move fast and the function is safety-relevant, cellular handover is designed for moving clients in a way WiFi roaming is not; the connection follows the truck without renegotiation. Where the fleet works indoors and out, one network covers the aisles, the yard, the trailer park and the gatehouse, the outdoor half being where WiFi economics collapse. And where the plant intends to run more than forklifts on the network, AMRs, cameras, condition sensors, the SIM-based security model and a single managed radio layer beat an accumulation of vendor meshes, each with its own controller and nobody responsible for the whole.
A 40,000 square-metre plant with an active yard typically needs a handful of small cells for full coverage, against several dozen access points for the equivalent WiFi footprint outdoors, and that arithmetic is usually what settles it.
Where should a plant manager start?
Not with a radio purchase. Start with HSG136 and an honest segregation audit, because barriers and walkways remain the highest-value spend on most sites, and because the audit's residual risks, the crossings and shared aisles that cannot be designed out, are precisely the places a proximity or RTLS system should cover. Fit the no-infrastructure tiers fleet-wide in the meantime; there is no reason to wait on lights and impact logging.
Then, before piloting anything networked, survey the site's radio environment, aisles, yard and trailer park together, and put the coverage requirement in the safety system tender rather than discovering it afterwards. Vendors quote confidently for the aisle they demonstrated in; the tender should make them quote for the yard in the rain. The rollouts that stall rarely stall at the sensors. They stall at the network, and that part is fixable before the first tag is ordered.