Your Safety Vest Makes You Invisible to the Car That’s Supposed to Save You

Two federal agencies walk into a crosswalk. OSHA says the construction worker standing in it must wear reflective clothing so drivers can see him. NHTSA says the car approaching him must have automatic emergency braking so it stops before hitting him. The IIHS just tested what happens when those two mandates meet.^[1] The car plowed through the worker at full speed.

Not a software glitch. Not a sensor malfunction. The reflective strips that make a human instantly recognizable to another human’s eyes — retroreflective tape articulating limbs and joints, the same pattern on every highway worker’s vest from Bakersfield to Bangor. They actively confounded the pedestrian detection algorithms. The AEB didn’t slow down because it couldn’t figure out what it was looking at.

IIHS tested three 2023 models: a Honda CR-V, a Mazda CX-5, and a Subaru Forester. The CR-V hit the test dummy in 84% of runs. The CX-5 hit it in 88%. When the dummy wore reflective strips on its limbs and joints (the standard configuration for ANSI Class 2 and 3 safety garments), neither the Honda nor the Mazda slowed at all.^[1] Not by 5 mph. Not by 1. The brakes never engaged.

The Subaru stopped every time but once. That one failure? The dummy was wearing reflective strips under 10 lux of roadway lighting.^[1] Even the system that works has a hi-vis blind spot.

The Scale of the Problem

This would be an academic curiosity if pedestrians weren’t dying at 40-year highs. In 2024, drivers struck and killed 7,148 pedestrians in the United States.^[2] Year-over-year it was actually a 4.3% decline, which the GHSA framed as progress. It’s still nearly 20% above 2016 levels.

Three-quarters of those deaths happened after dark.^[2] Fatal pedestrian crashes at night rose 84% between 2010 and 2023, while daytime fatalities increased just 28%. Nighttime is where the bodies are, which is precisely why NHTSA’s AEB mandate, finalized in April 2024 and requiring all new vehicles to have pedestrian AEB by September 2029, specifically requires nighttime detection capability.^[3]

NHTSA estimates the mandate will save 360 lives per year.^[3] That estimate assumes the technology works. Prior IIHS research already showed that pedestrian AEB’s real-world effect on crash risk in darkness is “negligible.”^[4] Negligible isn’t zero, but it isn’t 360 either.

The Regulatory Contradiction

Count the federal requirements in play. The FHWA’s Manual on Uniform Traffic Control Devices requires high-visibility apparel for all workers within the highway right-of-way. OSHA’s construction standards demand retroreflective garments in traffic exposure zones. ANSI/ISEA 107-2020 specifies exactly how those garments must be designed, including retroreflective strips placed on the torso, limbs, and joints to create the “biomotion” pattern that allows humans to instantly recognize a moving person.^[5]

That biomotion pattern is the problem. IIHS senior research scientist David Kidd put it plainly: “The placement and motion of reflective strips on the joints and limbs of pants and jackets allows drivers to quickly recognize the pattern of movement as a person. Unfortunately, the moving strips didn’t have the same effect for the pedestrian AEB systems we tested and probably confounded their sensors.”^[1]

What the human visual system reads as “that’s a person walking,” the camera-and-radar stack reads as noise. Decades of occupational safety research perfected these garments for one kind of perception system, and nobody checked whether they’d work for the replacement.

Who’s Standing in the Road

The Bureau of Labor Statistics counts roughly 3.5 million construction workers in the U.S., a significant fraction of whom work roadside in mandated hi-vis gear. Add utility crews, emergency responders, law enforcement directing traffic, tow truck operators. The FHWA estimates approximately 130 roadway workers are killed by vehicles each year. That number has stubbornly refused to drop despite improved signage, barriers, and increasingly fluorescent vests.^[6]

The AEB mandate was supposed to help. If two-thirds of tested systems can’t see the clothing these workers are legally required to wear, it won’t.

The Counterargument

Three vehicles is a small sample. The Subaru proved the problem is solvable — Forester’s system handled every scenario including reflective clothing with only one partial failure. Automakers are already improving nighttime AEB performance in response to IIHS ratings that now weight nighttime testing more heavily. The NHTSA mandate doesn’t take full effect until September 2029, giving engineers three more years to fix the sensor stack. By the time AEB is universal, this particular blind spot may be history.

That’s a reasonable position. It’s also the same argument the industry made about ESC in 2007. It took until 2024 for the fleet to fully turn over.^[7] Roadway workers dying between now and “eventually” will not find the timeline reassuring.

Limitations

This analysis rests on a three-vehicle IIHS study, nowhere near a representative sample. The test used black garments with reflective strips rather than full ANSI Class 2/3 hi-vis gear (fluorescent orange or yellow), which may behave differently under sensor scrutiny. All tests were conducted at 25 mph; real-world work zone crashes happen at higher speeds where AEB is even less effective. We also lack data on what fraction of the 5,300+ annual nighttime pedestrian fatalities involve workers in hi-vis versus civilians in dark clothing. The 130 worker deaths annually is a lower bound on exposure, not a comprehensive measure of the affected population.

What It Means

The federal government has spent fifty years perfecting the science of making humans visible to other humans. It now has three years to make sure the robots got the memo. The IIHS data suggests at least some of them didn’t — and the clothing that represents a half-century of occupational safety engineering may be actively working against the next generation of crash prevention technology.

Somebody should probably tell OSHA.

Source: IIHS pedestrian AEB study (January 2025), GHSA pedestrian fatality data (2024), NHTSA AEB final rule (FMVSS No. 127). This article examines the interaction between hi-vis safety apparel and automated emergency braking systems based on published research. The three-vehicle IIHS sample is small; results may not generalize to all AEB systems. See methodology for caveats.