Blog

The car plant and autonomy, written plainly.

Short pieces from the Chassira team on the body shop, the paint shop, torque, end of line and what it takes to let software move a setpoint in a car plant. Four so far. Each one is drawn from the thinking in our own planning documents, not from customer work, because there is none yet.

4 postsChassira team2026-10-09

Latest

A car every sixty seconds, built blind

Automakers ship around 90 million vehicles a year, each of roughly 30,000 parts, to millimetre geometry and Newton-metre torque, on lines that launch a car about every minute. Most of that is still judged after the fact.

2026-10-09 · 5 min read

2026-10-09 · 5 min read · Chassira team

A car every sixty seconds, built blind

Automakers ship around 90 million vehicles a year, each of roughly 30,000 parts, to millimetre geometry and Newton-metre torque, on lines that launch a car about every minute. Most of that is still judged after the fact.

A vehicle is stamped, welded, framed, painted, married to its powertrain or pack, trimmed, fastened and proven at end of line against a hard takt and a hard launch date. Each step is safety critical. A body-shop geometry drift, a paint-film defect, a mis-torqued fastener or a mis-calibrated ADAS sensor triggers rework, a line stop that costs tens of thousands of dollars a minute, a warranty claim or a recall that runs to billions.

Plants run this on sampled inspection, andon reaction, siloed cell PLCs and the tribal knowledge of launch engineers. Point tools exist for every step: robot lines, weld and paint controllers, press monitoring, in-line vision and metrology, torque traceability, battery and ADAS test stands, MES dashboards, twin authoring tools. Each owns a slice. None closes the loop from sensing a problem to acting on it, and none sees body, paint, powertrain and final as one chain.

The idea behind Chassira is that they are one chain. Stamping decides framing, framing decides gap-and-flush, the booth decides the repaint, the trace decides the warranty claim. Chassira fuses the signals, predicts the defect and acts inside approved limits, before the shift where it can and in station where it cannot.

2026-10-09 · 4 min read · Chassira team

Autonomy is a dial, and a weld schedule is the last thing it turns

A car plant does not go from manual to unattended in one step, and the crash-safety joints should not be the first thing an agent moves.

Design partners start in shadow mode. Chassira connects read-only to weld traces, metrology, vision, torque, test, robot and MES signals, measures the baseline for rework, downtime, first-time-through and warranty, and writes nothing. In assist mode it recommends weld, framing, paint, routing and sequencing moves, each with source evidence, confidence, the qualified envelope and a rollback path, and a process engineer approves each one.

Bounded autonomy follows for the low-risk loops: booth environment, framing offsets inside the envelope, blank feed and lubrication, routing, holds and sequencing, inside PLC and safety limits. Weld-schedule adaptation comes last, supervised, for qualified joint families only, after audited ROI, golden-dataset evaluation, twin validation and quality sign-off. Torque specifications, battery-joint parameters and end-of-line test verdicts never leave human hands.

Every action carries its evidence, envelope, confidence, estimated quality, rework, downtime and warranty impact, rollback path and audit entry. That format is what makes autonomy acceptable to plant, quality, STA and OT teams, and it is what makes trust possible before autonomy.

2026-10-09 · 4 min read · Chassira team

Hit build quality before the shift

Plants commit a shift of bodies to a weld sequence, a framing setup and a booth recipe before they know the result. The twin changes the order.

Whether a body hits geometry, a finish is defect free, a pack is torqued to spec and a line holds takt depends on a coupled chain from stamping to end of line. Today that chain is tuned by trial builds, offline simulation and launch craft, and launch ramps burn bodies and weeks while quality and takt collapse under new-model, EV and mixed-material complexity.

The Vehicle-and-Line Twin simulates welding, framing geometry, paint, torque and takt on an Omniverse plant twin, compares as-built against as-designed, and ranks candidate weld sequences, framing offsets, booth windows and line balances before the shift. The chosen setpoints go to the control agents, are executed inside their limits and approvals, and every real outcome recalibrates the twin.

Weld-thermal and geometry FEA, stamping and springback simulation, crash and structural surrogates, paint-flow simulation and point-cloud processing are compute-heavy workloads. Running them fast enough to inform a shift is the reason the twin is built on GPUs, and the reason the architecture notes that BMW and Mercedes-Benz already run production-scale Omniverse plant twins.

2026-10-09 · 3 min read · Chassira team

The launch engineer is leaving

The process engineers and technicians who hold a launch together cannot be hired fast enough for the EV and localisation wave.

Ask a plant manager and the same losses come up: weld and geometry rework per body, repaint and scrap in the paint shop, torque and fill faults escaping to end of line, battery-joint and ADAS faults escaping to the field, and launch ramps where quality and takt collapse together. The knowledge that prevents these lives in a handful of launch engineers, and simultaneous new-model, EV and battery-line launches are spreading them thinner every year.

Chassira captures that craft as a model: the full chain of panel dimension, weld trace, framing measurement, paint reading, torque trace, test result and disposition, per vehicle, with every supervised correction feeding the next shift. The plant runs at expert judgement on every shift, and the record of how is written into the vehicle's as-built record.

Run timeline · scenario

The scenario the posts refer to

Every Chassira run is an ordered, inspectable sequence. This is run_41a9c2 on Plant BIW-2 · EV crossover body shop to end of line · shift 1. Scenario completes in 10 steps with one approval gate. Weld schedule change on robot W-14 approved by the body-shop launch engineer, framing offset held inside the envelope, booth humidity trimmed before the predicted dirt defects appeared, one pack held at marriage on a torque trace, and the shift released with full as-built records.

run_41a9c2 Plant BIW-2 · EV crossover body shop to end of line · shift 1 illustrative scenario SUCCEEDED
  1. 01 ingest.shift SUCCEEDED 0.8 s

    Plant Orchestrator · Pulled the shift build sequence, program control plan and qualified envelopes from MES: 412 bodies, four trim variants, weld schedule set W-rev 7, framing targets per station, paint colour blocks, torque specifications. Locked every safety-critical parameter as read-only.

  2. 02 twin.simulate SUCCEEDED 1 m 40 s

    Vehicle-and-Line Twin · Simulated 18 candidate weld-sequence, framing-offset and booth-window combinations on the plant twin, ranked on predicted body geometry, expulsion risk at robot W-14, paint defect map and takt against launch cadence.

  3. 03 stamp.form SUCCEEDED continuous

    Stamp-and-Form · Blank feed on press line P-3 adjusted by 0.4 mm and lubricant dose raised 3 percent for the new coil lot as springback risk on the rear door inner climbed. Panel dimension held inside the window; genealogy handed to framing.

  4. 04 approve.human APPROVAL 3 m 12 s

    Plant Orchestrator · Candidate #6 lowers weld current 2 percent and adds 20 ms hold on robot W-14 to cut predicted expulsion. Weld schedules are crash-safety critical and stay with a human, so the change was held for the body-shop launch engineer and approved at 06:14:52.

  5. 05 weld.join SUCCEEDED continuous

    Weld-and-Join · Approved schedule 7b applied on W-14. Current and force traces confirmed inside the window on the first 40 bodies; predicted expulsion risk fell from 0.31 to 0.06 and no geometry debt was introduced at the adjacent seams.

  6. 06 frame.geometry SUCCEEDED continuous

    Frame-and-Geometry · Framing station FS-1 offset corrected by 0.3 mm on the left B-pillar datum inside the qualified envelope as in-line measurement trended toward the tolerance edge. Gap-and-flush vision confirmed the correction on the next 25 bodies.

  7. 07 paint.finish SUCCEEDED continuous

    Paint-and-Finish · Booth B-2 humidity trimmed 2 points and airflow rebalanced for the dark-colour block as the defect model predicted dirt on horizontal panels. Film build held inside the window on the first colour block.

  8. 08 marriage.torque SUCCEEDED continuous

    Powertrain-and-Battery · Pack PK-1182 held at marriage after a torque-to-yield trace on fastener 7 fell outside the specification curve. Routed to rework with the trace attached; torque specification untouched, as always.

  9. 09 inspect.fuse SUCCEEDED continuous

    Quality-and-Anomaly · Gap-and-flush and paint-defect cameras, weld and torque traces and test results fused per body. Two paint dirt defects localised to booth B-2 zone 3 before the humidity trim took effect; both routed to spot repair, none escaped to final.

  10. 10 eol.release SUCCEEDED 2 m 05 s

    EOL-and-Calibrate · End of line proved alignment, ADAS calibration, leak, brake, roll and functional tests on every vehicle; two ADAS retests sequenced inside takt. Shift 1 released with as-built records, approvals and dispositions written for the quality lead.

Autonomy

Four levels, set per agent and per tag

A car plant does not go from manual to unattended in one step. Chassira makes the level explicit, auditable and reversible at any time. The first release runs shadow mode, then assist mode, then bounded autonomy on low-risk loops. Weld schedules stay supervised until the joint family is qualified, and torque specifications and test verdicts are never written by an agent.

Autonomy levels and the human role at each
LevelWhat the agent doesWhat the human doesWhere it sits in a pilot
L1 · ShadowReads weld, metrology, vision, torque, test, robot and MES signals; predicts and recommends; writes nothingRuns the line as today; compares predictions to audit and engineer outcomesDesign partners start here; the baseline is measured from the plant's own records
L2 · AssistProposes weld, framing, paint, routing, holding and sequencing moves with evidence, confidence, envelope and rollback pathApproves each action in the review consoleAssist mode: accuracy and ROI proven against the signed baseline
L3 · BoundedExecutes pre-approved low-risk actions: booth environment, framing offsets, blank feed and lubrication, routing, holds, sequencing, inside PLC and safety limits and the qualified envelopeApproves weld-schedule changes, torque-related decisions and vehicle releaseGraduated autonomy, after audited accuracy and ROI
L4 · ScaledSupervised weld-schedule adaptation for qualified joint families, across lines and plantsSets the envelope; reviews the shift record and evaluation resultsPlanned; after golden-dataset evaluation, twin validation and quality sign-off. Torque specifications and test verdicts never leave human hands
Accelerated computing

Physical AI, at the line

A spot weld lasts a fraction of a second and a body passes a station every minute. Perception has to be local, deterministic and fast, so Chassira runs GPU inference at the plant edge and keeps training, simulation and optimisation in the cloud or on-prem.

Jetson Orin · DeepStream · Holoscan · TensorRT

Edge perception

8 to 32 cameras per process area at 15 to 60 FPS, plus torque, dimensional, thermal, robot and paint-instrument streams fused at the cell. Weld and gap alerts under 50 ms and paint or dimensional prompts under 150 ms are the design targets. Jetson Thor is planned for higher density.

Triton · NIM

Model serving

Weld-defect, geometry-risk, paint-risk, torque and battery, and takt-risk models served at the plant edge with versioned rollback, routed by latency, tenant and cost, with canary rollout by line.

DGX or HGX · NeMo

Training

Weld, paint and gap vision, dimensional models, NeMo IATF and spec reasoning and GPU-accelerated stamping, springback and weld-geometry FEA surrogates. 150 to 600 GPU-hours per plant during onboarding, then monthly retraining from engineer corrections. Dedicated capacity is planned.

Omniverse · OVX

Plant twin

Body framing, weld sequence, robot reach, paint flow and line balance simulated before the shift. The pattern is production proven at BMW and Mercedes-Benz, per the architecture; Chassira's twin and OVX capacity are planned.

Cosmos · Omniverse Replicator

Synthetic data

Rare weld defects, paint failures, geometry-drift modes, occluded joints and misbuild cases generated and validated against real distributions before promotion. 50,000 to 250,000 variants per program is the design target. Planned.

cuOpt

Optimisation

Line sequencing and JIS feeding, AGV routing, takt, line balancing, buffers and retest sequencing.

Isaac · Isaac ROS

Robot cells and AMRs

Robotic weld, paint, assembly and inspection cells and AGV and AMR line-feed autonomy, with skills validated in simulation before a cell runs them.

RAPIDS

GPU ETL

Telemetry ETL across weld and press parameters, measurements, paint readings, torque traces, battery and ADAS results, non-conformance and takt history for retraining and benchmarks.

NeMo Guardrails · NVIDIA AI Enterprise

Policy and deployment

Spec, IATF, functional-safety, cyber and human-approval policy enforced before any write; supportable on-prem or plant-edge deployment for regulated OEM plants.

Who it is for

Built for the people who run the plant

We have no customer quotes to show yet and will not invent any. These are the three roles in a car plant we build for, and the problems they bring to the first meeting.

Plant Manager or Manufacturing Director

Owns launch dates, takt, downtime, first-time-through, rework and cost.

What this person is dealing with today

  • Sampled inspection, andon reaction and siloed cell PLCs that let joint quality, geometry, paint and torque drift
  • Launch ramps where quality and takt collapse under new-model, EV and mixed-material complexity
  • A skilled process-engineer and technician shortage that cannot staff the localisation and EV launch wave

Body, Paint or Assembly Process (Launch) Engineer

Owns weld, paint and torque process windows, framing geometry, dimensional accuracy and launch quality.

What this person is dealing with today

  • Weld, rivet and adhesive parameter and sequence drift driving crash-safety joint defects, geometry loss and rework
  • Body-in-white dimensional and gap-and-flush error at framing costing downtime and warranty per body
  • Paint dirt, runs, orange peel and e-coat and film-build faults driving repaint, scrap and corrosion warranty

Quality, STA or Conformance Lead

Owns build quality, PPM, warranty and recall exposure, and IATF 16949, APQP and functional-safety traceability.

What this person is dealing with today

  • Torque, fluid-fill, battery-joint and ADAS-calibration faults escaping to the field as warranty and recall exposure
  • Stamping, body, paint, powertrain, battery, assembly and end of line siloed and reactive across the plant
  • Traceability evidence assembled by hand after the fact
FAQ

Straight answers

The questions plant managers, launch engineers and quality leads ask in the first meeting.

Yes, but only within an explicit tag allow-list with per-tag rate and magnitude limits, inside the qualified envelope, and only at the autonomy level your plant has set. Design partners start in shadow mode, where Chassira writes nothing. Weld-schedule changes wait for a named body-shop engineer, and torque specifications, battery-joint parameters and end-of-line test limits and verdicts are read-only to every agent, always.

Get started

Talk to the people who wrote this

Start with one process and one measurable baseline. A paid pilot on body-shop weld and gap-and-flush vision, paint-defect detection and film-build control, or torque and battery-joint conformance shows the number before you commit further.