Walk onto any active civil construction site and you will hear it within seconds. Reversing alarms, movement alarms, radio calls and shouted instructions, often several at once, layered over plant noise, generators and traffic. It is a constant soundscape, and that is precisely the problem.
Audible alarms are one of the oldest and most widely used controls on civil sites as they are cheap, simple and have been standard practice for decades. But a control that has been standard for decades is not automatically a control that still works as intended, and constant, undifferentiated noise is losing its ability to warn anyone of anything.
The stakes are not theoretical. Safe Work Australia recorded 188 worker deaths from traumatic injuries in 2024. At least one vehicle was directly involved in 66 per cent of them, and construction accounted for 37 of those deaths, one in five of the national total (Safe Work Australia, 2025).
Why the Alarm Itself Is Part of the Problem
Alarm fatigue is a well-documented human factors phenomenon, not a discipline problem. When a warning sound is present almost continuously in someone's working environment, the brain does what it is designed to do with any constant, low-value stimulus. It filters it out.
How this plays out on a busy civil site:
- Workers form a learned association between the alarm and routine, harmless movement, because most of the time that is exactly what it is
- The alarm carries no information about direction, distance or urgency, only that something, somewhere, is moving
- Over a full shift, sustained noise exposure further reduces a worker's capacity to register and respond to any single alert
But research into standard tonal reversing alarms shows the fault is not only behavioural. When the Noise Regulation Branch of Western Australia's Department of Environment and Conservation measured tonal and broadband reversing alarms across 26 government and council vehicles, it found the pattern of sound behind a vehicle carrying a tonal alarm is very uneven, with variation of up to 8 decibels at a single measuring position seven metres directly behind the machine, so a worker can be clearly in the alert zone one moment and effectively out of range the next, without moving at all (Popoff-Asotoff, Holgate and Macpherson, 2012).
SafeWork NSW's Code of Practice for moving plant on construction sites adds that when multiple machines carry identical alarms, the resulting confusion can make the warning devices "ineffective" (SafeWork NSW, 2025).
The Code's own suggested remedy points to combination audio-visual systems and broadband, or white noise, alarms where site noise levels vary. WorkSafe WA's guidance on audible reversing alarms goes further, advising any organisation that relies on audible reversing alarms for safety to consider broadband (WorkSafe WA). Broadband alarms are a real improvement on the pure tone, and on many sites the sensible first step, but they do not change the underlying architecture. The alarm still sounds every time the vehicle moves, still broadcasts to everyone within earshot whether or not anyone is at risk, and still depends on a worker hearing it, interpreting it and reacting correctly. That is the behavioural ceiling administrative and passive controls reach on active sites: a control technically present and functioning as designed, doing very little in the moment it matters.
Spotters Face the Same Ceiling
Trained spotters are often positioned as the answer to this gap, and a competent spotter giving close attention to a single manoeuvre is genuinely effective. But they cannot be everywhere a heavy vehicle moves, cannot hold that attention across a full shift, and are themselves a person on foot in the vehicle's path if a miscommunication occurs. SafeWork NSW's Code of Practice reflects this directly, requiring that businesses eliminate or minimise the need for spotters to be in shared zones or near moving plant, so far as reasonably practicable, rather than treating a spotter as a default solution (SafeWork NSW, 2025). Adding more people to manage a visibility gap does not remove the gap. It adds another person exposed to it.
Where Active Detection Changes the Equation
Radar and camera detection systems address the weakness audible alarms and spotters cannot overcome: they do not depend on a sound field reaching and being correctly located by a worker, and they respond to a person or object entering a defined zone around the vehicle rather than sounding continuously regardless of whether anyone is nearby. This aligns with the general direction of the SafeWork NSW Code, which requires higher-order engineering and isolation controls to be used unless it can be shown this is not reasonably practicable, with procedural measures such as alarms and spotters treated as supporting controls rather than the primary line of defence (SafeWork NSW, 2025).
The shift that matters is from broadcasting to targeting. Zone-based detection stays silent until someone is actually within the vehicle's danger zone, and it alerts the operator, the one person who can stop the machine, rather than the fifty people who cannot. Because the alert is rare, it keeps the meaning a continuous alarm loses.
Two questions usually follow. The first is whether in-cab detection simply relocates alarm fatigue to the operator. It will, if zones are set too wide or the system is not configured for the working environment, which makes commissioning and zone tuning part of the control rather than an afterthought. The second is cost across a mixed-age fleet, and retrofit is rarely a whole-of-fleet decision. It is usually a decision about which vehicles, in which environments, carry the highest exposure, and starting there.
Procurement Is Already Moving This Way
This is not only a safety argument. It is increasingly a procurement one.
CLOCS-A, the national Construction Logistics and Community Safety Australia standard, sets out tiered accreditation across Bronze, Silver and Gold, and the direction of the tiers is instructive. Bronze establishes a baseline that pairs the audible warning with reversing sensors or cameras. Silver adds left-side blind spot detection and left-turn warnings. Gold adds telematics monitoring. Each step moves further from a vehicle that simply announces itself and closer to one that detects what is actually around it.
CLOCS-A is a voluntary standard, but that distinction is narrowing in practice. Transport for NSW projects now require Silver as a baseline for heavy vehicles on their contracts, and both Sydney Metro and Melbourne Metro have written elements of the approach into theirs. For contractors bidding into the infrastructure pipeline, the practical question is less whether the standard applies and more which tier their fleet can currently evidence.
The standard and the human factors research arrive at the same place from different directions. One reflects what procurement is beginning to require. The other explains why.
A Practical Review Worth Running
Fleet and safety teams reviewing civil construction operations could start by asking one question while standing on an active site: if that alarm sounded right now for a genuine reason, would anyone notice? On many sites, the honest answer points toward active, targeted detection as the next layer of control, sitting alongside spotters and procedures rather than replacing them. A second question follows naturally: if the same site were assessed against CLOCS-A tomorrow, which tier would the fleet be able to evidence?
Safety and risk teams wanting to assess where administrative controls on their sites may be reaching this ceiling can refer to the Heavy Vehicle Safety Requirements Information Pack, which sets out common operational risks and example control measures.