Industries / Transport and vehicles / Automotive / Towing
Safety at the roadside
Roadside safety is keeping tow operators, patrols and stranded drivers from being hit by passing traffic while they work.
7
Milestones achieved, from 1996 to 2025.
1
To watch for: announced, not yet confirmed outside.
3
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
1996
The first Move Over law
On January 28, 1994, paramedic James D. Garcia was struck by a car going about 45 mph while checking on a driver near Lexington, South Carolina. In 1996 South Carolina passed what the National Sheriffs' Association calls the first US Move Over law, which requires drivers to take care when passing emergency scenes; the Emergency Responder Safety Institute (ERSI) later said it "sounded nice but had no teeth" and was rarely enforced. By 2012 all 50 states had similar laws, by the sheriffs' account.
LessonA law that drivers do not know and police do not enforce protects few people.
Credit
James D. Garcia; South Carolina General Assembly
2012
Responders train together
The second Strategic Highway Research Program (SHRP2) produced the National Traffic Incident Management (TIM) Responder Training in 2012, a 4-hour course that teaches police, fire crews, medics, road agency staff and tow operators together to clear crashes quickly and safely. By October 2025 more than 800,000 responders had taken it, by the Federal Highway Administration's count, against a goal of 1.3 million.
Credit
SHRP2; Federal Highway Administration (FHWA)
2019
Towing deaths come into view
A study of US Occupational Safety and Health Administration (OSHA) files found 106 cases of tow operators killed or badly hurt while helping motorists; nine in ten were struck, mostly by traffic or rolling vehicles, or caught between vehicles or in moving parts. It cited federal counts of 231 deaths in the towing industry from 2011 to 2017, 37% of them operators struck by traffic, and the authors' earlier figure of 54.6 deaths per 100,000 towing workers in 2015 against 3.4 for all industry. It also noted that employers need not promptly report a death on a public road to OSHA.
Credit
Mark D. Chandler and Terry L. Bunn, Kentucky Injury Prevention and Research Center
2021
Navigation alerts measured
From May to July 2021, researchers led by Darcy Bullock of Purdue University studied queue warning trucks on Indiana interstates, which park ahead of stopped traffic with lights, message boards and digital alerts sent to drivers' navigation apps. Using hard-braking records from connected cars over about 370 hours with the trucks and 58 hours without, they found about 80% fewer hard-braking events with the trucks. The study measured lights, signs and alerts together, not the alert alone. Alerts of this kind are carried by services such as HAAS Alert's Safety Cloud, which sent its first alert on June 1, 2016, and by its own count passed 2 billion alerts in October 2022, the 2 billionth from a tow company.
Credit
Rahul Sakhare, Darcy Bullock and colleagues; HAAS Alert
2023
England drops new smart motorways
On April 17, 2023, the UK government cancelled all plans for new smart motorways, roads where the hard shoulder becomes a traffic lane, citing a lack of public confidence and cost pressures; 14 schemes were dropped. It said stopped-vehicle detection, which spots a car halted in a live lane, had been installed on every all-lane-running smart motorway, and that 150 more emergency areas would be built by 2025.
LessonTaking away the place where broken-down cars stop lost public confidence even with detection systems added.
Credit
UK Department for Transport; National Highways
2024
Missing from crash data
The AAA Foundation for Traffic Safety identified 123 roadside assistance providers, including tow operators, mobile mechanics and road agency patrols, struck and killed at the roadside in the US from 2015 to 2021: slightly more than the federal workplace death count and far more than national crash data showed. Crash records usually list a worker struck outside a vehicle as a pedestrian. More than half were hit on the shoulder or beyond by a driver who had left the road, nearly two-thirds died in darkness, and records of lights and safety clothing were missing in most cases.
LessonWorkers on foot are recorded as pedestrians, so national crash counts miss many roadside worker deaths.
Credit
Brian C. Tefft, Anqi Wei and Rebecca Steinbach, AAA Foundation for Traffic Safety
2025
Cameras check who moves over
Using traffic-camera video and software that tracks each vehicle, the AAA Foundation measured 12,365 drivers passing 169 roadside incident scenes in 13 states. 64% changed lanes or slowed and 36% did neither; when only a tow truck or a road agency patrol was present, 58% did, against 66% when only police were there, and few slowed by as much as state laws required. ERSI counted 12 tow operators among 46 responders killed by passing vehicles in 2024, and 11 among 36 in 2025.
newsroom.aaa.comnewsroom.aaa.comlearning.respondersafety.com
Credit
AAA Foundation for Traffic Safety (Benedick, De Leonardis and others); ERSI
Watch for
Alerts that slow drivers
HAAS Alert says drivers who receive its Safety Cloud alerts slow down by an average of 17%. Kodiak, a driverless-truck company, says its trucks send these alerts automatically when stopped on the shoulder, reaching drivers through car screens and apps including Waze and Apple Maps.
Achieved whenan independent study measures how much drivers slow at tow or roadside-assistance scenes when alerts are sent, against matched scenes without alerts, and publishes the difference.
Claimed by
HAAS Alert and Kodiak, February 2026
Open
01
Alerts tested at tow scenes
A test of digital alerts at real tow and roadside-assistance scenes: how long after the truck's lights go on the alert reaches Waze, Apple Maps and connected cars; how far up the road drivers get it; and whether passing drivers change lanes or slow more than at matched scenes without alerts. Everything is recorded from inside vehicles or from traffic cameras, so nobody stands at the roadside for the study.
The briefCredit · unclaimed02
Case files for roadside deaths
A public case file for each tow operator and roadside worker killed by passing traffic in the US since 2015, coding the scene: time and light, road type, speed limit, where the worker stood, the truck's light colors, cones or flares, high-visibility clothing, whether a police or patrol vehicle was blocking, and whether the striking driver left the road.
The briefCredit · unclaimed03
Light colors on trial
A comparison of how drivers pass tow trucks showing amber lights only and tow trucks also showing blue or red, using recorded traffic-camera video from states with different rules, measured the same way across hundreds of scenes.
The briefCredit · unclaimed
Open 01 · the brief
Alerts tested at tow scenes
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A test of digital alerts at real tow and roadside-assistance scenes: how long after the truck's lights go on the alert reaches Waze, Apple Maps and connected cars; how far up the road drivers get it; and whether passing drivers change lanes or slow more than at matched scenes without alerts. Everything is recorded from inside vehicles or from traffic cameras, so nobody stands at the roadside for the study.
Why it is still open
The two independent field studies found did not test tow trucks. A February 2026 University of Illinois study for the state transportation department filmed thousands of vehicles from a helicopter passing a stopped police car and a state maintenance truck in 2024 and 2025: alerts moved lane changes and slowing farther upstream, and drivers got them about half a mile ahead, but at the time they worked fully only in Waze. Purdue's 2021 Indiana study covered queue trucks that combine lights, signs and alerts. HAAS Alert's 17% speed figure is its own.
Built with what exists
Phones running navigation apps as receivers in a car driving the route; recorded video from state traffic cameras; open-source vehicle detection and tracking (for example Ultralytics YOLO with ByteTrack); the counting method of the AAA Foundation's 2025 study.
First step
Partner with one tow company that already sends alerts and log, from its dispatch records and a following car's phones, the alert delay at five scenes.
Done looks like
At least 40 scenes, half with alerts, with the time from lights-on to alert for each app, the share of passing vehicles that moved over or slowed, and a published comparison with its margin of error.
Open 02 · the brief
Case files for roadside deaths
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A public case file for each tow operator and roadside worker killed by passing traffic in the US since 2015, coding the scene: time and light, road type, speed limit, where the worker stood, the truck's light colors, cones or flares, high-visibility clothing, whether a police or patrol vehicle was blocking, and whether the striking driver left the road.
Why it is still open
The AAA Foundation's 2024 study found that records of safety clothing, warning lights and other equipment were missing in most of its 123 cases, and OSHA need not be told promptly of deaths on public roads. ERSI lists each death with a news link but does not code the scene.
info.oregon.aaa.comstacks.cdc.govlearning.respondersafety.com
Built with what exists
ERSI's yearly lists, the federal FARS fatal crash files, police crash reports obtained by public-records request and NIOSH FACE investigation reports; an AI model that pulls the same fields from each document (see the Note "The jobs agents repeat"), with every field linked to the sentence it came from (see the Note "Findings that cite notes") and checked by a person.
First step
Code the 23 tow-operator deaths on ERSI's 2024 and 2025 lists from their linked news stories, and request the police reports.
Done looks like
At least 250 coded cases with 12 or more fields each, two coders agreeing on at least 90% of fields in a 50-case sample, and a published table of how often each scene factor appears.
Open 03 · the brief
Light colors on trial
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A comparison of how drivers pass tow trucks showing amber lights only and tow trucks also showing blue or red, using recorded traffic-camera video from states with different rules, measured the same way across hundreds of scenes.
Why it is still open
The AAA Foundation's April 2025 review of warning-light research found that it "does not clearly point to the best lighting solutions for safety" and asked for long-term studies on tow and roadside service vehicles, since most past work used boxy trucks such as fire engines. Its October 2025 video study compared kinds of responders, not light colors.
Built with what exists
Recorded video from state traffic management centers; the state-by-state light rules in the AAA Foundation review; open-source vehicle detection and tracking (Ultralytics YOLO, ByteTrack); a statistics notebook.
First step
Use the review's state tables to pick two neighboring states with different tow-light rules, and ask their traffic centers for recorded video of tow scenes.
Done looks like
At least 300 tow scenes from at least four states, at least 100 of them with blue or red lights, with the share of drivers who moved over or slowed under each light scheme, by day and by night, published with the code.