In short: A container terminal is one of the few workplaces where 60-tonne machines and people on foot share the same ground at speed, often in the dark and the rain. The technologies that prevent the worst outcomes — proximity warning, geofencing, real-time location — are only as good as the wireless layer underneath them, and on most UK quays that layer is the weak link.
Key Takeaways
- Segregation breaks down at the interface — keeping people and machines apart works until the lashing gangs, drivers and surveyors have to enter the stack, and it is exactly at that interface where proximity warning has to take over.
- Proximity systems need real-time location to work — collision-avoidance is a latency problem; a warning that arrives a second late on a contended network is a warning that arrives too late.
- One network beats a box per machine — bolting an isolated sensor onto every vehicle creates islands of data, whereas a single private 5G network turns every machine and worker into a node in one live operational picture.
In a nutshell
Ask anyone who has run a container terminal what keeps them awake, and the answer is rarely throughput. It is the thought of a straddle carrier and a person occupying the same patch of tarmac at the same moment. A modern terminal is a choreography of very large machines — straddle carriers, reach stackers, rubber-tyred gantries, terminal tractors, internal movement vehicles — moving heavy steel boxes around the clock, in weather, across a site where visibility from a cab perched several metres up is poor by design. Into that environment walk lashing gangs, hauliers, surveyors, fitters and stevedores, on foot, because the job cannot be done any other way. The Health and Safety Executive's docks guidance and the long-standing Docks Regulations exist because this combination has, repeatedly and predictably, killed people.
This post is about the part of port operations that does not appear in the throughput statistics: keeping the people and the machines apart, and what happens at the points where they cannot be kept apart. It is, more than most people realise, a connectivity problem.
Segregation is the rule, and the rule has gaps
The foundation of port safety is segregation — designing the operation so that pedestrians and vehicles simply do not share space. Marked walkways, barriers, separate access routes, exclusion zones around working cranes, controlled crossing points. Where it can be achieved, segregation is the most effective control there is, because it removes the hazard rather than managing it. Good terminals invest heavily in it and Port Skills and Safety, the sector's safety body, has pushed it hard for years.
The trouble is that segregation is never complete, because the work itself requires people to enter the machines' space. Containers have to be lashed and unlashed by hand on the quay. Drivers have to get out at the interchange. Surveyors have to inspect. Fitters have to recover a broken-down machine where it stopped. At every one of these points, a person on foot is deliberately inside the operating area of equipment that cannot see them well and cannot stop quickly. The HSE's incident record for the sector is dominated not by exotic failures but by this mundane, recurring scenario: someone struck by a reversing or turning vehicle at the interface where segregation runs out.
The honest position is that you cannot engineer the people out of a working quay. So the safety case has to include a second layer that operates precisely where segregation fails: technology that knows where the machines are, where the people are, and warns both before they meet.
Proximity warning is a latency problem in disguise
Proximity warning and collision-avoidance systems have matured a great deal. A worker wears a tag; a machine carries a detector; when the two come within a defined distance the driver gets an alert and, increasingly, the machine itself can be slowed or stopped. Vehicle-to-vehicle versions warn drivers of each other around blind corners and between container rows. The better systems fuse several technologies — radio ranging, ultra-wideband, sometimes camera-based detection — to cut false alarms, because a system that cries wolf gets switched off or ignored, and an ignored safety system is worse than none.
What is easy to miss is that all of this is fundamentally about time. A straddle carrier travelling across a yard covers a surprising distance in the fraction of a second it takes a warning to be generated, transmitted, received and acted upon. Collision-avoidance only works if that whole loop closes fast and, critically, reliably — every time, not on average. A proximity alert that is correct but arrives a beat late because the network was busy is not a safety system; it is a record of an accident that was about to happen. This is why the wireless layer underneath matters so much, and why it cannot be a best-effort, contended one.
Contended Wi-Fi across a large outdoor yard is poorly suited to this. Coverage is patchy between the steel canyons of stacked containers, handover as a machine moves between access points is exactly where packets get lost, and there is no mechanism to guarantee that a safety message takes priority over a routine data transfer. Private 5G changes the equation: engineered coverage across the whole yard from a small number of radios, clean handover built into the cellular design, and — through network slicing and quality-of-service — the ability to put safety-critical proximity and location traffic on a guaranteed path ahead of everything else on the network.
Real-time location is the common foundation
Proximity warning between a tag and a vehicle is the sharp end, but the more powerful idea is a terminal-wide real-time location system: a continuously updated picture of where every machine, every vehicle and, where the workforce agrees, every person is across the whole site. Once that picture exists and is trustworthy, a great deal becomes possible. Dynamic geofencing can create soft exclusion zones around a working crane that move with the operation and shrink the area where people and machines mix. The control room can see a lashing gang has entered a stack and automatically slow nearby movements. A near-miss can be reconstructed from real data rather than conflicting accounts. Speed limits can be enforced by zone.
None of this works on snapshots. It needs a constant, low-latency stream of position updates from a large number of moving assets simultaneously — the massive-device-density, low-latency profile that cellular technology was specifically designed for and that Wi-Fi was not. A real-time location system is, in effect, a digital twin of the safety situation, and like any digital twin it is only as good as the freshness and completeness of the data feeding it. Stale data is dangerous data when the thing being tracked weighs sixty tonnes.
The trap of a box per machine
When a terminal first tackles this, the instinct is often to buy a self-contained safety product — a sensor-and-alarm unit, one per machine, working in isolation. It is an understandable first step and it does save lives. But it creates a familiar problem: islands. Each machine's system knows only what its own sensors see. The data does not aggregate into a site picture. Adding a capability means touching every machine again. And the safety estate ends up as a patchwork of vendor boxes that do not talk to each other, with no single operational view.
The alternative is to treat connectivity as the shared foundation and the safety applications as services that ride on it. With one private 5G network across the terminal, every machine and every worn device becomes a node in a single live picture. Proximity warning, real-time location, geofencing, speed monitoring and incident reconstruction stop being five separate products and become five views of the same underlying data. New capabilities are deployed to the network rather than installed machine by machine. This is the same argument that drives the move away from per-vehicle, per-camera point solutions across the whole smart-port story — and on safety, where the cost of fragmentation is measured in response time, it is at its most compelling.
The honest limits
We should be straight about what a network does and does not do, because over-claiming on safety is how trust gets lost. A private 5G network does not, by itself, prevent a single collision. It is enabling infrastructure: it makes proximity warning fast enough to be useful, makes terminal-wide location possible, and makes a unified safety picture practical. The controls that actually stop accidents — the segregation design, the warning systems, the procedures, the training, the culture — still have to be there and still have to be good. A network laid over a poor safety culture buys very little.
There is also a workforce dimension that has to be handled with care and not engineered around. Real-time tracking of people is a sensitive matter, and rightly subject to consultation, agreement and clear limits on how the data is used. The ports that have made progress here have done so by being explicit that location data exists to keep people alive at the vehicle interface, not to monitor productivity — and by keeping the two firmly separate. That is a matter of trust as much as technology, and the technology choice does not absolve anyone of getting the trust right.
What we conclude
A container terminal is one of the most dangerous workplaces in ordinary industrial life precisely because it cannot fully separate people from very large, fast, heavy machines. Segregation is the first and best control, but it always has gaps at the points where the work forces people into the machines' space, and it is those gaps that the sector's incident record keeps filling. The technologies that address them — proximity warning, geofencing, real-time location — are real and maturing, but every one of them is a latency-and-coverage problem before it is anything else.
We conclude that vehicle-pedestrian safety on the quay is, at its foundation, a network question. UK port operators — from the ABP and Forth Ports estates to the big container terminals — already run the machines and increasingly run the sensors. What ties them into a safety system that responds in time, rather than a collection of isolated alarms, is a single, deterministic wireless layer beneath them. On the most dangerous fifty metres of the working quay, that layer is the difference between a warning that arrives in time and one that merely records what happened.