Why unwanted alerts undermine proximity safety and how configurability fixes it
A proximity system that alerts too often doesn't stay switched on for long. This is one of the best-documented patterns in safety science, and it applies just as much to a forklift on a warehouse floor as it does to a hospital ward. Here's what the evidence actually shows, and why configurable zones — not fewer sensors — are the real fix.
85–99%
Of clinical alarm signals found not to require intervention, per the Joint Commission
26
Successful mobile-plant prosecutions in Victoria in 2025, second only to falls from height
4
Distinct vehicle operating states SonaSafe's zones automatically reconfigure for
The Real Cost
Unwanted alerts don't just annoy people. They get switched off.
WorkSafe Victoria issued a safety alert after an employee was hit and injured by a reversing excavator. The machine had a motion alarm fitted, exactly as it should. It had also been switched off, via a cab-mounted switch, because the alarm's constant noise had become a nuisance in built-up areas. WorkSafe's investigation found this wasn't an isolated case: a broader pattern of powered mobile plant being fitted with switches specifically to silence warning devices operators had grown tired of.
That's the real cost of an unwanted alert. It isn't irritation. It's a safety-critical device being disabled by the very people it was installed to protect, because the system gave them a reason to stop trusting it.
A Well-Documented Pattern
This isn't unique to proximity systems — and that's useful to know
Healthcare has studied this exact problem in more depth than almost any other industry, because the stakes of getting it wrong are so immediate. The Joint Commission, the body that sets US hospital safety standards, has found that 85 to 99% of clinical alarm signals don't actually require intervention. The consequence is well documented too: clinicians exposed to that volume of unwanted alarms respond more slowly, and in some recorded cases stop responding to genuine alerts at all.
Human-factors research on sustained attention backs this up from a different angle. Foundational vigilance research by Wickens and Hollands, along with Bliss, Gilson and Deaton's work on alarm effectiveness, shows that repeated false alarms measurably reduce how quickly and reliably people respond to real ones. Mica Endsley's situation awareness research adds the mechanism: sustained attention depends on trusting that what you're perceiving is meaningful, and a system that regularly cries wolf erodes exactly that trust.
Why It's An Engineering Problem
A control that gets switched off has already stopped working
An engineering control that fires so often it gets disabled has effectively demoted itself to no control at all — whatever the specification sheet says it's capable of. WorkSafe Western Australia's own guidance on reversing alarms makes a related point clearly: the answer to a nuisance alarm isn't to remove it, it's to make it smarter, because a safe system of work still needs the warning in place. That's a useful distinction. The problem was never that the alert existed. It's that it wasn't targeted enough to earn ongoing trust.
The Real Fix
Configurability: matching the alert to the actual risk
A fixed-size detection zone treats every moment the same way, regardless of what the vehicle is actually doing. That's the single biggest driver of unwanted alerts, and it's also the easiest thing to fix properly, provided the underlying system is built to be shaped rather than just switched on. SonaSafe's zones adjust automatically to what the vehicle is doing, not just where it is.
Zone extends at the rear
The direction of travel is where the risk actually is, so that's where the zone grows — rather than maintaining a symmetrical bubble that alerts on people who were never in the vehicle's path.
Wedge-shaped zone, extended ahead, narrowed at the sides
At travelling speed the real risk is what's ahead, not what's beside the vehicle. Narrowing the sides cuts unwanted alerts from people and objects the vehicle was never going to reach.
Zone extends at front and sides
Low forward speed usually means loading or close manoeuvring — precisely when pedestrian risk is highest. The zone widens here deliberately, trading a slightly higher alert rate for protection exactly when it matters most.
GPS-defined areas where alerts are suppressed
Where a physical barrier or another control already manages the risk, there's no reason for the system to keep alerting on it. Geo-zones let a site switch alerts off exactly where the risk genuinely doesn't exist, without weakening coverage anywhere else.
Different rules for different trained roles
A hatchman working close beside a crane on a container terminal is doing his job correctly, not creating a hazard — but a generic proximity rule would alert on him constantly. Tags can be configured to each role's actual working distance, so a trained person doing close-proximity work by design isn't drowned in alerts that don't reflect real risk.
None of this is about making the system less sensitive. It's about making every alert mean something, which is the entire difference between a system people trust and one they quietly work around.
Why The Underlying Technology Matters
Not every configuration has the same unwanted-alert profile
This is also where the choice between Tagged, Tagless and Hybrid genuinely matters, not just for detection reliability but for trust over time. A UWB-only system without a second sensing layer tends to alert against a fixed zone shape, which is exactly the pattern that produces nuisance alerts in busy areas. A camera-only system can misidentify non-human objects as people in the wrong lighting or conditions. Pairing sonar with UWB is what allows the zone itself to be shaped — by direction, by speed, by role — rather than firing on anything that crosses a fixed radius, which is the mechanism behind everything described above.
Proven In Practice
This is exactly what a pilot is for
Mainstream NZ's 30-day pilot didn't just prove SonaSafe could detect people reliably. It proved the system had been tuned to the site's actual traffic patterns before it went live — and the result was an 80% reduction in near-misses and a 65% reduction in exclusion-zone breaches, numbers that only hold up if the alerts behind them were trusted, not tuned out. Getting zone shapes, speed thresholds and safe zones right isn't a settings menu nobody opens. It's the specific work a properly run pilot exists to do.
THE PATTERN
What this means in practice
Trust is a finite resource, and every unwanted alert spends it
Whether it's a hospital ward or a container yard, the mechanism is identical: enough unwanted alerts, and people stop believing the next one is real.
The fix is precision, not volume
Removing an alarm because it's annoying trades a known nuisance for an unknown risk. Shaping the alert to match the actual hazard solves the real problem instead.
Configurability is what keeps an engineering control switched on
A control that gets disabled provides no protection at all. The system has to earn ongoing trust to keep doing its job.
Sources
- WorkSafe Victoria — Warning devices on powered mobile plant, safety alert
- WorkSafe Western Australia — Audible reversing alarms: considerations for use
- WorkSafe Victoria — 2025 penalties summary
- The Joint Commission — National Patient Safety Goal on clinical alarm system safety (NPSG.06.01.01), via Seifert et al., 2021
- Wickens, C. D., & Hollands, J. G. — Engineering Psychology and Human Performance
- Bliss, J. P., Gilson, R. D., & Deaton, J. E. — Human performance and alarm effectiveness in a vigilance task
- Endsley, M. R. (1995) — Toward a Theory of Situation Awareness in Dynamic Systems
How many of your alerts are actually being trusted?
A site assessment maps how your zones should be shaped — by direction, speed and role — before a single unwanted alert has the chance to erode trust in the system.
Talk to our team
