Radar Alignment Station Software for ADAS Assembly Lines
Radar alignment station software for ADAS assembly lines: vision-aligned mounting, per-fastener torque capture, alignment re-verified after fastening, a sealed module self-test, and a signed per-unit alignment report.
Schematic
Radar aiming and interlock
How boresight is measured against a target, and how the result gates the build.
Radar alignment verification that checks where the beam points
MileSoft ADAS Radar Assembly is an end-of-line alignment and validation station for automotive radar modules feeding adaptive cruise control, lane keeping and automatic emergency braking. A vision-aligned fixture positions the module, torque and angle are captured per fastener on a multi-vendor controller backbone, alignment is re-verified after fastening, and the module runs a sealed self-test against a fixed station fiducial to report where its beam actually points.
The fixture does not release a unit until alignment is within tolerance and the self-test has passed, and every unit ships with a signed report carrying serial or VIN, azimuth, elevation and roll, torque per fastener, station, operator and timestamp, exportable to the customer's traceability format. Station calibration is validated on a schedule an external auditor will recognise.
This is alignment verification on the assembly line, not workshop ADAS recalibration. In a published Tier-1 automotive deployment integrated with the plant's manufacturing execution system, the station is credited with a 92% reduction in customer containment incidents.
Radar alignment station
A radar module can sit correctly and point somewhere else
Vision measures where the housing is. The vehicle cares where the beam points.
Radar modules feeding adaptive cruise control, lane keeping and automatic emergency braking are unforgiving. Sub-degree misalignment passes visual inspection, so nothing at the plant notices, and it degrades a function drivers exercise rarely — so it surfaces months later as a customer complaint and a containment letter. A vision system can be arbitrarily precise about where the housing sits and still observe nothing about where the antenna beam goes. The offset between the two, the boresight offset, lives inside the individual module.
| Term | Origin | Observed by |
|---|---|---|
| Placement error | Fixture and operator placement | Vision, before fastening |
| Fastening shift | The tightening sequence | Vision after fastening — and nothing earlier |
| Boresight offset | Inside the module | The functional self-test only |
| Measurement error | The station's own instruments | Nothing — absorbed by guard banding |
That is why each module boots a sealed self-test against a fixed station fiducial and reports its own measured alignment. It is not a more accurate dimensional check; it is the only check of a different kind. A station that tightens its vision tolerance is buying precision on three terms of a four-term budget.
Post-fastening verification
Why radar alignment is re-checked after the screws are driven
Fastening is a perturbation, not a fixation — even when the tightening is perfect.
A tightening sequence applies torque about an axis and draws a joint down unevenly as fasteners take load, so it moves the part. Torque and angle are captured per fastener on the same multi-vendor controller backbone as torque traceability — VDI/VDE 2862 puts a mounting whose misalignment degrades braking at the top of its classification. But correct torque and unchanged position are independent facts, so a station that verifies only before fastening certifies one position and ships another.
Worked example: a healthy fastening process still creates escapes
Tolerance 0.30°. Placement error σ = 0.09°; fastening shift centred at 0.04° with σ = 0.07°. A unit measured at 0.26° before fastening is a routine pass — and exceeds 0.30° afterwards with probability about 0.72. Across the placement distribution, about 3.1% of units passing the first check fail after fastening. Without the post-fastening check, all of them ship, and none looks wrong at any point.
The fastening process here is entirely healthy. The escapes come from its interaction with placement — which is why tightening more carefully does not fix them, and why the post-fastening check is the one most often cut on cost grounds and the one that must not be.
Boresight offset
The alignment term no camera sees
A station that verifies only dimensionally is wrong about its own capability, in the optimistic direction.
Worked example: adding the term inside the module
Placement and fastening together give a spread of 0.114°. Add a boresight offset distribution with σ = 0.11° and it becomes 0.158°; with 0.04° of station measurement uncertainty, 0.163°. Against a 0.30° tolerance, the dimensional view suggests a comfortable 2.6σ; the functional reality is 1.8σ. The station is wrong about its own capability by 39%, and no amount of vision resolution moves the second number.
Variant control upstream matters here too. A close-cousin radar variant has a different nominal boresight relationship, so a wrong-variant module can pass every dimensional check and fail functionally — which is why this station is paired with pick-to-light kitting upstream.
Pass verdict
A pass verdict needs a guard band, an interlock and a measured record
For a safety-related part, the evidence has to be per unit and it has to be measured.
acceptance limit = specification limit − k × combined measurement uncertainty
For a safety-related part the choice of k is not symmetric: a false reject costs a rework at the station, an escape costs a containment programme. The uncertainty cannot be assumed; it comes from a repeatability and reproducibility study on the station's own instruments (ISO 5725, AIAG MSA).
A radar feeding emergency braking is a safety-related item under ISO 26262, whose Part 7 places obligations on production itself: the process must be capable, must detect when it is not, and must demonstrate both afterwards. So the fixture does not release the unit until alignment is in tolerance and the self-test has passed — a control function whose release circuit is designed under ISO 13849-1, because a lock that fails open lets an unverified safety part onto a vehicle.
Every unit ships with a signed report: serial or VIN, azimuth, elevation and roll, torque per fastener, station, operator and timestamp, exportable to the customer's traceability format. It is worth something in a dispute only if the alignment values are the self-test's measurements rather than the fixture's targets, the operator is authenticated, and the station produced it at the time. The station's own fiducial is validated on a schedule — and because every unit passed since the last validation is suspect if it drifts, the calibration interval is a containment-breadth decision, not a maintenance one.
Drift detection
Detect the drift, not the defect
An escape costs every unit built while nobody knew, so the lever is how early the process is seen to move.
population at risk = production rate × time from onset to detection
Worked example: a worn fixture on a 5,000-module-a-month line
The alignment mean walks toward tolerance; yield does not change, because the drift sits inside the guard band for six weeks. The customer raises a containment letter four months after onset: 20,000 modules at risk, and at a containment cost of 45 each, 900,000 plus programme costs. A drift alarm on the alignment distribution firing at week two — mean moved one standard deviation, every unit still passing — puts 2,500 at risk, most still in the plant or in transit.
The alarm did not get better at catching a bad unit; it detected a moving process. That is worth 17,500 modules of exposure, and it is the only lever that acts on the multiplier rather than the per-unit cost. Track alignment mean and spread per station and shift, alarm on movement rather than failure, and treat a shifted distribution as an investigation even when yield is unchanged.
Field evidence
The published result, and the caveats that belong with it
One attributed deployment result, reported carefully.
A Tier-1 automotive supplier building radar modules found off-spec alignment slipping through quality control and surfacing at the customer. After integrating this station with its manufacturing execution system — each radar booting a sealed alignment routine and shipping a signed report — the deployment reports a 92% reduction in customer containment incidents, attributed to this station.
Incidents are lumpy
A count over an unstated period; one containment programme can span thousands of units, so fewer incidents is not proportionally fewer units affected.
No baseline rate published
The account gives the reduction, not the starting incident rate.
Kitting changed at the same time
The same programme installed pick-to-light upstream, which prevents wrong-variant modules reaching the station. Its zero wrong-part picks and five-day ramp-up belong to that module, not this one.
The account is in the pokayoke case study. No numeric alignment tolerance is published in MileSoft's material; "sub-degree" describes the order of magnitude of the problem, not a specification.
Evaluating alignment stations
Eight questions to put to any radar alignment station
Each is answerable by demonstration — including by us.
| # | Dimension | Ask the vendor to demonstrate |
|---|---|---|
| D1 | Post-fastening check | Is alignment re-measured after the screws are driven, or only before? |
| D2 | Functional measurement | Does the module report its own alignment, or does the station infer it? |
| D3 | Release interlock | Fail the self-test deliberately. Does the fixture let the unit go? |
| D4 | Failure direction | If the release circuit fails, does the fixture hold or open? |
| D5 | Stated uncertainty | What is the station's measurement uncertainty, and how was it established? |
| D6 | Guard band | Is the acceptance limit tighter than the specification limit, and by how much? |
| D7 | Measured pedigree | Are the values on the per-unit report measurements, or the targets? |
| D8 | Drift alarm | Does the station alarm on a distribution shift, or only on a failing unit? |
D8 is the question that acts on cost rather than defects. Every other dimension improves the chance of catching a bad unit; only this one shortens the interval in which bad units are being built.
Standards referenced: ISO 26262-1 and -7, IATF 16949, VDI/VDE 2862 Part 1, ISO 6789-1 and -2, ISO 22514-1 and -2, ISO 5725-1 and -2, AIAG MSA, ISO 13849-1. Derivations are in the ADAS radar assembly method paper.
Features
Key Features
Built-in capabilities that set our software apart
Vision-aligned fixture
the unit is not released until alignment is within tolerance and the self-test passes
Closed-loop torque
torque and angle captured per fastener on the same multi-vendor controller backbone as torque traceability
Post-fastening re-verification
alignment measured again after the screws are driven
Sealed self-test
the radar reports its own measured alignment against a fixed station fiducial
Signed per-unit report
serial or VIN, azimuth, elevation, roll, torque per fastener, station, operator and timestamp
Station calibration validation
the fiducial and instruments validated on an auditable schedule
Alignment drift alarms
mean and spread tracked per station and shift, alarming on movement rather than failure
Benefits
Key Benefits
What your team gains from adopting our platform
Testimonials
What Our Clients Say
Trusted by industry leaders across manufacturing, HVAC, and logistics.
"Our “Part-Process Matrix” as a concept and Software received 1st runner-up award in MVML Innovation fair 2014. As a customer we appreciate your support and efforts to deliver it on time and most importantly as per customer requirement."
FAQ
Radar Alignment Station Software — FAQs
Answers to the most common questions about this product. Need more? Book a demo or reach out via Contact.
Why do standard assembly fixtures fail for radar modules?
Why re-check alignment after fastening?
What is a sealed self-test routine?
Do we get per-unit alignment reports?
Which standards frame this?
Explore More
Related Products
Complementary tools teams often pair with this solution
Related Reading
Go deeper
Explore the long-form story behind this product.
Dimensionally Correct Is Not Functionally Aligned: Fastening Shift, Boresight Offset, and the Self-Test That Closes the Loop
Architecture, Computational Methods, and Field Evidence from MileSoft ADAS Radar Assembly
Read the paperRead the full Radar Alignment Station Software for ADAS Assembly Lines deep-dive
Read articleSee how MileSoft delivers in Pokayoke
Read case studySee your savings in under a minute
Estimate the time and cost this tool could give back — free, instant, no sign-up.
See the check a camera cannot make
Bring your alignment tolerance and fastening data. We will show how many passing units a post-fastening check would catch, and what boresight offset adds to your capability figure.