Industries / Transport and vehicles / Automotive / Collision repair
Sensors and calibration
Many cars now carry cameras, radars and ultrasonic sensors for driver-assistance systems (ADAS), such as automatic emergency braking (AEB); after a crash or a windshield change, shops scan the car's computers for faults and calibrate the sensors, re-aiming them to the carmaker's tolerances with fixed targets or a test drive.
6
Milestones achieved, from 2016 to 2026.
3
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
2016
Glass code from insurers' lab
In July 2016 the ADAS Repair Group, led by Thatcham Research, the UK insurers' research center, and including the glass replacers National Windshields and Autoglass and the repairer Nationwide Crash Repair Centres, issued a voluntary ADAS Glazing Code of Practice on recalibrating sensors when a windshield is replaced or refitted. In May 2019 Thatcham widened its advice to any repair that touches ADAS sensors or the parts near them, and said calibration must be "undertaken and confirmed within vehicle manufacturer tolerances."
Credit
Thatcham Research and the ADAS Repair Group
2016
Scan every car
On 9 June 2016 FCA US, maker of Chrysler, Jeep, Dodge and Ram, told body shops and insurers that every vehicle must be scanned for diagnostic trouble codes, the fault codes stored by a car's computers, before and after every repair. It named glass replacement and even a disconnected battery as events that could set such codes, and listed forward cameras, radar and blind-spot monitors among the systems to check. It called its own wiTECH scan tool important, saying it held software that aftermarket tools "may not contain".
Credit
FCA US
2017
One degree delays braking
The Insurance Institute for Highway Safety (IIHS) tested factory and aftermarket replacement windshields on late-model Honda Civics. On one aftermarket windshield the camera bracket was off by one degree from the factory mount, and that was enough to make the automatic braking act later and the lane departure warning sound too early or too late.
LessonIIHS concluded that windshields with ADAS cameras should be calibrated whenever they are replaced, "whether they are OEM or aftermarket.".
Credit
IIHS (Sean O'Malley)
2018
Sensors double small repair bills
AAA, the US motoring club, priced repairs on three popular models and found in October 2018 that a minor front or rear collision on a car with ADAS could cost up to $5,300, almost two and a half times the bill for a car without them. Replacing a windshield with a camera behind it cost about $1,500, up to three times as much. AAA's 2023 update, on three 2023 models with parts priced at carmakers' list prices, put a front radar at $500 to $1,300 and moving and calibrating a windshield camera at $360, on top of the body work.
Credit
AAA
2025
Car data law survives court
In 2020 Massachusetts voters approved, by about three to one, a law giving owners and independent shops access to the repair data that cars send by telematics (wireless links to the maker). Carmakers' trade group, the Alliance for Automotive Innovation, sued, and on 11 February 2025 US District Judge Denise Casper dismissed the suit, ruling that the law covers only data for diagnosis, maintenance and repair and that carmakers have ways to comply.
LessonRather than comply, Subaru and Kia left their telematics services off some 2022 and newer cars sold in Massachusetts, which the judge noted as one lawful route.
Credit
Massachusetts voters; Judge Denise Casper
2026
Calibration mostly sent out
Researchers at the University of North Florida surveyed 304 US collision repair businesses. Most were single-site shops with ten or fewer technicians; only 72 had technicians dedicated to ADAS, and 42 of those had formal training from a carmaker. Most shops ran pre- and post-repair scans themselves but sent calibration to outside specialists, and nearly two-thirds reported trouble hiring and keeping ADAS technicians or relied on such specialists instead.
Credit
Dongyuan Wang, Pingying Zhang and Xiaohan Wu
Watch for
Robots place the targets
For a static calibration the technician sets a target board at an exact distance, height and angle from the car. Autel says its IA1000 uses a robotic frame that places the target so that pitch, roll, yaw, height and distance are right every time, removing guesswork; the coverage gives no accuracy figures.
Achieved whenan outside test publishes placement error against a surveyed reference, and calibration pass rates, for robot and manual setups on the same cars.
Claimed by
Autel, at the NADA Show, 2026
Calibration as routine upkeep
Ascential Technologies, which sells calibration equipment through its Burke Porter brand, tested one 2024 Nissan Altima with about 13,000 miles under US government (NHTSA) test procedures. Untouched, the car failed to stop before the target in four of seven braking tests; with no calibration, lane departure warning failed completely; with a good calibration it performed as from the factory, and even a poor calibration beat the untouched car. Ascential says sensors drift in normal use and calibration should become routine maintenance. Its paper also cites Highway Loss Data Institute findings that Subarus and Hondas with ADAS kept lower crash-claim rates at five years and older.
Achieved whenan independent test of many used cars measures how far sensors drift with age and mileage.
Claimed by
Ascential Technologies, July 2025
Repair data in federal law
On 21 May 2026 the US House Energy and Commerce Committee voted 48 to 1 for the Motor Vehicle Modernization Act (H.R. 7389). Its section 202 would make law the 2014 national agreement, based on Massachusetts's repair law, under which carmakers give independent shops the same diagnostic and repair information they give dealers. Provisions of the broader REPAIR Act (H.R. 1566) on telematics and wireless access to car data were left out.
Achieved whenthe provision passes both houses of Congress and is signed into law.
Claimed by
House Energy and Commerce Committee, May 2026
Open
01
Radar through repaired bumpers
A public table of how much a car radar's signal, in the 76 to 81 GHz band most car radars use, loses passing through bumper covers repaired the way body shops repair them: filler over a dent, extra coats of paint, a metallic color over the factory one. Each cover is measured before and after repair.
The briefCredit · unclaimed02
Count the missing calibrations
An audit of finished repairs that lists, for each car, the calibrations its maker's repair procedures require for the work done, and checks them against the calibration reports and the final invoice to see which were done.
The briefCredit · unclaimed
Open 01 · the brief
Radar through repaired bumpers
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A public table of how much a car radar's signal, in the 76 to 81 GHz band most car radars use, loses passing through bumper covers repaired the way body shops repair them: filler over a dent, extra coats of paint, a metallic color over the factory one. Each cover is measured before and after repair.
Why it is still open
Published measurements use new or laboratory samples. Jennie Lodén's 2017 KTH master's thesis measured 25 samples from a Volvo Cars supplier, some repainted with two aftermarket basecoats, and found that more metal in the basecoat raised its permittivity (how strongly a material holds back and reflects radio waves). Expósito and colleagues (Sensors, 2023) measured bumper samples in an anechoic chamber, a room that absorbs stray radio echoes, and found losses below 1.25 dB, larger with thicker varnish. Neither tested covers repaired in shops. Rohde & Schwarz sells its QAR50 radome tester for design and production lines. A 2026 ATZ article on repairing bumper covers in front of radar sensors is behind a paywall and could not be read, and no public data on shop-repaired covers was found.
Built with what exists
matching used bumper covers from salvage yards; a 77 GHz radar evaluation board such as Texas Instruments' AWR1843BOOST with TI's free mmWave Studio software, aimed at a corner reflector (a metal target that returns a strong echo) at a fixed range, comparing the echo with and without the cover; or a vector network analyzer with horn antennas in a university lab; a coating thickness gauge for plastics; a body shop doing each repair to the paint maker's instructions.
First step
Buy two matching used covers, measure the echo through each on the bench, have one refinished with one extra coat, and measure again.
Done looks like
At least 30 covers from at least three models, each measured three times before and after one of at least four repair types, with the loss in dB and its uncertainty, and a second lab reproducing ten of the covers within 0.5 dB.
Open 02 · the brief
Count the missing calibrations
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
An audit of finished repairs that lists, for each car, the calibrations its maker's repair procedures require for the work done, and checks them against the calibration reports and the final invoice to see which were done.
Why it is still open
Caliber Collision, a US repair chain, says 65% of collision repairs needed an ADAS calibration by the second quarter of 2025, up from 53% at the end of 2024, from repairs analyzed with its partner Protech's ADAS ID3 tool. Claims data count the calibrations that appear on estimates (see Estimating), a much lower share, and the University of North Florida survey found most shops send calibration to outside specialists, so the work often leaves the shop that wrote the estimate. These figures count different things, and no public audit was found that matches required calibrations to completed ones car by car.
Built with what exists
carmakers' position statements and repair procedures; closed repair files from shops that agree to take part, with owner details removed; an AI model that answers fixed questions from each procedure, such as Google's Gemini Decision 1 (see the Note "A thousand questions, one pass"), with every answer checked by a person; NHTSA's free vPIC service, which decodes a VIN into make, model and year.
First step
Ask one shop for 20 closed files on cars from the last five model years and check them by hand against the makers' procedures.
Done looks like
At least 300 repairs from at least five shops, each with required and completed calibrations listed and sourced to a page of the procedure, the share of required calibrations not done with a 95% confidence interval, and a second reviewer agreeing on at least 90% of cars.