transport · family: the missing yardstick
calibrated, but is it right?
After a repair, shops have no settled way to prove the car's crash-avoidance sensors work as designed again — about half of owners who had these features repaired report problems afterward
Problem statement
Automatic emergency braking, lane keeping and blind-spot warning depend on cameras and sensors that must be precisely calibrated; trade reporting says a sensor pointing just two degrees off center line may cause the safety systems to behave as though the vehicle were in an entirely different environment. Windshield replacement often makes calibration necessary, IIHS describes calibration as a common step in many repairs, and automakers stipulate it anytime a sensor is removed and replaced or reinstalled — but once the procedure is done, confirming that the system actually performs to specification remains, in the trade press's words, "largely unsolved." A clean scan showing no fault codes is not confirmation that the sensors are operating within specification, and most vehicles have no instrumentation that tells a driver or a technician whether they are. Two executives quoted in the same reporting say that primary responsibility for the work stays with the shop that accepted the vehicle, even when the calibration is sublet.
Why this matters
The IIHS surveyed nearly 500 drivers who had had a crash-avoidance feature repaired and found that about half reported problems with the feature afterward; among owners whose repair involved windshield replacement the figure was about two-thirds, and among those repaired for crash damage nearly three-quarters — and about two-thirds of respondents said calibration had been part of the repair, so calibration is being done and yet the systems still misbehave. As IIHS's Alexandra Mueller put it, "Many had issues with the technology afterward, and some said they had to have the same feature repaired more than once." In trade-press reporting from the collision industry in 2026, the chief executive of Repairify estimates that roughly 65 percent of vehicles going through collision shops need some calibration and that only 35–40 percent of that work is actually getting done. Because IIHS research shows these features substantially reduce the crashes they are designed to address, a silently mis-aimed sensor returns a car to the road with a safety system that looks intact but isn't.
What’s been tried and why it hasn’t worked
Manufacturers publish calibration procedures, but IIHS notes that the processes are not standardized and that some calibrations are complicated and require large spaces, specialized training and expensive equipment; and the procedures describe how to run a calibration without stating which sensors a given pattern of collision damage should trigger — and, by one remote-diagnostics provider's estimate, 40–50 percent of the calibrations it flags to shops are ones the shop is hearing about for the first time. One shop owner quoted by Autobody News calls it one of the industry's biggest misconceptions that ADAS decisions should be based on whether a warning light is on the dashboard; as the same owner says, "The absence of a warning light does not mean the system is operating as designed"; a code can sit in pending status until the vehicle completes enough of a drive cycle to set it, and sensors also drift over time without any collision. IIHS itself identified what is missing: simplified, standardized procedures across manufacturers, adequate technical information for repair shops, a centralized database of calibration specifications, and self-diagnosing systems that alert a driver or technician when something is wrong. Federal legislation (the ADAS Functionality and Integrity Act, folded into H.R. 7389 in 2026) would, as Autobody News describes it, direct NHTSA to establish testing guidelines and tolerances; the executive director of the Society of Collision Repair Specialists reads the bill as focused more on understanding how vehicle modifications affect these systems than on establishing accountability mechanisms to ensure that repair and calibration procedures are actually performed.
What would unlock progress
The missing piece is an independent, manufacturer-agnostic post-repair functional verification — a way for a shop to demonstrate that a calibrated camera or radar produces the correct detection ranges and lane geometry, analogous to how a wheel-alignment rack reports measured toe and camber against a spec rather than trusting that the technician followed the procedure. Adjacent precedents suggested by this write-up, not drawn from the two sources: end-of-line ADAS verification in vehicle assembly plants, aviation's post-maintenance functional checks, and metrology practice where a calibration is only complete when a traceable measurement confirms it. A second unlock is a damage-to-sensor decision aid that maps repair operations to the sensors whose alignment they can disturb; Autobody News reports that the industry has made progress on this identification side and that identification systems are working on the mapping, while the remote-diagnostics estimate above indicates many calibration needs still go unidentified.
Entry points for student teams
A team with a used ADAS-equipped vehicle (or a bench camera module) could build and validate a low-cost verification protocol — controlled target boards at measured distances plus logged sensor outputs — and quantify how much aim error a shop-floor calibration leaves behind under different floor and lighting conditions. A software team could prototype the damage-to-sensor decision aid from published repair procedures and collision-estimate line items. A policy/human-factors team could design the pass/fail report an insurer and owner could both trust, drawing on IIHS's list of missing elements. Skills: mechanical/mechatronic engineering, computer vision, metrology, automotive service knowledge.
Genes
Genes marked “+” were added in a later review.
Structural cousins: stuck the same way, in other fields
Sources
"Crash avoidance features improve safety but complicate repairs," Insurance Institute for Highway Safety (IIHS), 2023-02-15, accessed 2026-08-17; "ADAS Calibration in 2026: The Gap Between Getting It Done and Getting It Done Right," Lisa McArdle, Autobody News, 2026-06-09, accessed 2026-08-17 go to source 1 ↗ go to source 2 ↗
Working notes
Notes made while writing and checking this brief, shown as written. They use working shorthand.
IIHS is the insurer-funded research institute whose survey data (HLDI Bulletins Vol. 36 No. 35 and Vol. 37 No. 7 are cited on the IIHS page) make this a tier-2 analyst source; the 2026 Autobody News piece is trade press quoting collision-industry practitioners (SCRS, CIC) and is used only for the practitioner framing and the 65% / 35–40% / 40–50% figures, which are industry estimates and should be treated as such by the verifier. IIHS's related finding that even about 0.6 degrees of camera misalignment materially delays automatic emergency braking surfaced in secondary coverage but was not read at the primary source and is deliberately left out of the body text. `failure:ignored-context` was chosen because manufacturer calibration procedures assume dealer-grade floors, lighting and information access that independent shops do not have (deployment/operational sub-pattern); `constraint:coordination` was considered and rejected — manufacturers do not agree on a common approach, so filter (1) fails. `temporal:newly-created` reflects that the problem exists only since ADAS proliferated on mass-market vehicles (roughly the last decade); worsening was rejected because the growing affected fleet is a growing-market pattern, not a hardening barrier. Related collection briefs: `transport-av-edge-case-detection-gap` (perception failures of autonomous systems, not post-repair calibration) and `transportation-automated-driving-monitoring`. Verified at intake 2026-08-17: gate (net) + adversarial source check + contested-tag second coding. Verifier note: all IIHS survey figures and quotes confirmed at the IIHS page; the 65% / 35–40% / 40–50% figures are single-executive estimates quoted in Autobody News (Repairify CEO; AirPro COO) and are hedged accordingly in the body.
Reconciliation 2026-10-01: The whole write-up was read against its two sources, both fetched in full on 2026-10-01 directly from the addresses on the Source line: the IIHS news article (https://www.iihs.org/news/detail/crash-avoidance-features-improve-safety-but-complicate-repairs) and the Autobody News article (https://www.autobodynews.com/news/adas-calibration-in-2026-the-gap-between-getting-it-done-and-getting-it-done-right). (1) The long title changed. Old: "After a Collision Repair, Nobody Can Prove the Car's Crash-Avoidance Cameras See Straight Again — Half of Owners Report the Features Still Misbehave." IIHS's "about half" is of owners who had a feature repaired for any reason (for crash damage it is nearly three-quarters), and Autobody News says verification is "largely unsolved," not that nobody can do it. New: "After a Repair, Shops Have No Settled Way to Prove the Car's Crash-Avoidance Sensors Work as Designed Again — About Half of Owners Who Had These Features Repaired Report Problems Afterward." The Display title is unchanged. (2) The text said sensor aim "is fixed to fractions of a degree." Neither source says this; Autobody News says a sensor two degrees off center line may cause the systems to misbehave, and the sentence now says that. (3) "Suspension work or bumper swap," and calibration "against manufacturer targets," are not in either source. The sentence now gives what IIHS says: windshield replacement often makes calibration necessary, calibration is a common step in many repairs, and automakers stipulate it whenever a sensor is removed and replaced or reinstalled. (4) "A repair shop has no independent way to prove" was stronger than the sources, in which one shop describes using manufacturer tests and road tests; the sentence now uses the Autobody News wording that verification remains "largely unsolved." (5) The statement that the only signals are the scan tool reporting the routine "completed" and the absence of a fault code, and the pedestrian-braking example, are not in the sources. The sentence now says what Autobody News says: a clean scan is not confirmation that sensors are within specification, and most vehicles have no instrumentation that shows whether they are. (6) "Repairers, insurers and vehicle owners are left arguing over work that no one can verify" is not in either source; it was replaced by the point, made by two executives quoted in the article, that primary responsibility stays with the shop even when calibration is sublet. (7) The 65 percent and 35–40 percent figures were presented as the trade press's own; they are one executive's estimate (the chief executive of Repairify) of vehicles "going through collision shops," and the sentence now says so. (8) The description of manufacturer procedures as static and dynamic, proprietary, and differing in target geometry, floor levelness and lighting is not in either source. The sentence now gives IIHS's points: no standardization, and some calibrations need large spaces, specialized training and expensive equipment. (9) "The industry has leaned on diagnostic trouble codes as proof of success" is not a statement either source makes; the sentence now gives only what the quoted shop owner says, that basing decisions on a warning light is one of the industry's biggest misconceptions; the drive-cycle clause now follows the article's account of pending codes. (10) IIHS's self-diagnosing systems alert "when something is wrong"; the text had "when a sensor is out of tolerance." (11) The sentence on SAE J3262 (an Information Report approved December 2023, a terminology standard, a reporting framework in development) is not in either source and was removed. (12) "Precisely because none exist today" was removed from the legislation sentence, and the article's qualification from the executive director of the Society of Collision Repair Specialists was added (the bill focuses more on understanding modifications than on accountability mechanisms for repair and calibration). (13) The precedents in the unlock section (assembly-plant checks, aviation, metrology) are now labeled as this write-up's own suggestions, and the decision-aid sentence now notes the article's report that the industry has made progress on identification and that identification systems are working on that mapping, alongside the remote-diagnostics estimate that many calibration needs still go unidentified. Confirmed in the sources and left as written: both citations on the Source line (titles, dates 2023-02-15 and 2026-06-09, author Lisa McArdle); nearly 500 drivers; about half reporting problems; about two-thirds for windshield replacement and nearly three-quarters for crash damage; about two-thirds saying calibration was included; the Alexandra Mueller quotation and the shop owner's quotation, both word for word; the 40–50 percent estimate from the remote-diagnostics provider; sensor drift; IIHS's list of what is missing; H.R. 7389; the two HLDI bulletins named above. Not checked: the 0.6-degree finding mentioned above, which the notes already say was not read at its source and which is not in either article. Two phrases in the notes above are not in the sources and should be read as the tagger's reasoning: that procedures assume "dealer-grade floors, lighting and information access," and "roughly the last decade." A second reader went over these repairs against both articles and the two HLDI bulletins on the same day; the wording of items 3, 5, 6, 9, 12 and 13 was tightened so that each sentence says no more than the sources do. On that pass, item 6 was attributed to two executives (the chief executive of Repairify and the chief operating officer of AirPro Diagnostics) rather than to the executives quoted generally; item 9 now states the misconception as the shop owner states it, about "ADAS decisions" rather than calibration decisions; and the clause that joined the how-versus-which point to the 40–50 percent estimate no longer presents the second as a consequence of the first, which the article does not say. No repair was undone. Genome tags untouched.