Linamar Onsite Primer
Walking the plant floor on Tuesday 2026-05-19 — business, processes, vendors, vocabulary.
CAD $2.264B
Q1 2026 revenue
$758M
Q1 2026 new business wins
8.1%
Normalized op margin Q1 (vs 6.6% Q1-25)
~5,000
Industrial robots deployed (mostly ABB)
“In robotics, we've signed an LOI to be the contract manufacturer in North America for Cobots. We partner with two separate parties to build humanoids and are also working with software companies on artificial intelligence development for the brains of those humanoids.”
01Why this brief
You walk the floor Tuesday with Dana Sharp, Lina Qamar, Mackenzie Kuntz, and Tom Schuyt. They will assume you know the plant-floor language. This brief makes you fluent in: what their business is doing right now, how a Tier-1 powertrain plant actually runs, where Antioch slots into the bin-pick-and-place pilot, and what their twelve closest peers are publicly doing on robotics and AI.
The walking-tour spine takes you station-by-station through a plant you've never been inside. Every jargon term is defined where it first lands. The deployment-reality section covers the safety standards updated in 2025, ROI math, the integrator ecosystem, and the vision-vendor faceoff for bin-picking. The Tier-1 analogues section answers "who else is doing what you do" with primary sources, including the three peers with named humanoid programs. The cheat sheet at the end gives you the "if they say X, you respond Y" responses for each of the four named attendees.
The hero quote is your unlock. Jarrell's "we partner with two separate parties to build humanoids and are working with software companies on the brains" is the public version of what Tom told you privately as "we want to be a North American component manufacturer for someone." Cite the earnings-call quote on Tuesday. It legitimizes the latent component-manufacturer pitch and tells you the buying motion is now sanctioned at the CEO level.
02The Linamar picture
Canadian-headquartered diversified industrial manufacturer. Founded 1966 by Frank Hasenfratz. Two reporting segments since 2025: Mobility (auto/powertrain Tier-1) and Industrial (Skyjack aerial work platforms + Agriculture). Run on a propulsion-agnostic strategy with semi-autonomous plants.
Business architecture and 2025 splits
| Segment / Region | 2025 Sales (CAD) | 2025 Op Earnings |
|---|---|---|
| Mobility (auto / powertrain Tier-1: cylinder blocks + heads, gears, transmissions, driveline, knuckles, structural castings, battery enclosures) | $7.73B | $562.8M |
| Industrial (Skyjack MEWPs + Agriculture: MacDon, Salford, Bourgault) | $2.49B | $329.3M |
| Canada | $5.14B | |
| Rest of North America | $2.02B | |
| Europe | $2.30B | |
| Asia Pacific | $0.755B |
Customer concentration that drives risk
In 2025 the two largest Mobility customers were 20.1% and 16.5% of total revenue (combined ~36.6%). Linamar is the sole supplier worldwide for products that represent more than half of Mobility sales [L-10]. One launch slip is material. This is why "validate in sim before you touch the line" lands so cleanly with this customer.
Leadership voices
“They were all accretive right out of the gate. I mean, the assets were distressed, you know, we negotiated ahead of acquisition to make sure that they'd be accretive day one.”
“In robotics, we've signed an LOI to be the contract manufacturer in North America for Cobots. We partner with two separate parties to build humanoids and are also working with software companies on artificial intelligence development for the brains of those humanoids.”
“I think AI, distribution, mega centers are a big part of the market that we're supplying.”
“From a European and North American standpoint, we're the first one.”
Recent M&A that reshapes the parts mix
Linamar uses acquisitions to acquire specific casting and forging capabilities. Each one introduces parts that are excellent bin-picking pilot candidates because they are heavy, irregular, and recently arrived without an in-house ABB cell footprint yet.
| Acquisition | Closed | Price | What it brought | Pilot fit |
|---|---|---|---|---|
| Georg Fischer Leipzig [L-3][L-11] | Dec 2025 | €45M | ~300 employees; Europe's largest molding box for machine-molded iron castings; 3D-printed sand cores; won fully-machined heavy-duty truck axle program immediately post-close. | HIGH — heavy ductile-iron handling |
| Aludyne North American assets [L-4][L-12] | Nov 2025 | $300M USD | Lightweight aluminum chassis + structural (knuckles, subframes, control arms, axle housings); generated $250M+ additional opportunities within months. | HIGH — high-volume structural feed |
| Dura-Shiloh battery enclosures [L-13] | Aug 2023 | — | 3 facilities (North Macedonia, Czechia, Alabama); battery enclosures for BEVs; in Linamar Structures Operating Group. | MEDIUM — multi-material kitting |
| Bourgault [L-14] | Feb 2024 | — | Saskatchewan; ag air-seeding leader; contributed $370M sales / $11.5M net earnings Feb-Dec 2024. | LOW — less immediate bin-picking |
The Guelph footprint and where the pilot likely lives
Linamar's Guelph campus is the global HQ and houses multiple production and engineering facilities. Public disclosure surfaces three by name; others (Concentric Equity, Frank Hasenfratz Centre of Excellence, McLaren Engineering, Tribologix) appear in private/internal naming and don't show up in 2025-2026 filings.
| Facility | Address | Function | Pilot fit |
|---|---|---|---|
| Vehcom Manufacturing | 74 Campbell Rd, Guelph [L-6][L-7] | Automotive components; active Environmental Compliance Approval (renewed); operating Linamar nameplate. | HIGH — true production environment; template for Leipzig/Aludyne replication. |
| Linex Manufacturing | 355 Massey Rd, Guelph [L-8] | Manufacturing site | MEDIUM — pending material-flow verification onsite. |
| Corporate HQ / R&D | 287 Speedvale Ave W [L-9] | Global HQ | MEDIUM — exec visibility, not grit. |
Vehcom is the leading pilot candidate. When Dana proposes a plant, ask "would Vehcom be a fit?" — the bin-pick-and-place demo there generalizes cleanly to Aludyne casting (aluminum knuckles, subframes) and Leipzig (heavy iron) for Phases 2-3.
Strategic priorities, next 3 years
- Mobility: propulsion-agnostic structural components to hedge EV softness; M&A as the capability-acquisition vehicle. Q1 2026 NBWs span ICE (cylinder blocks + heads), structural (knuckles), and CV (HD truck axles via Leipzig).
- Skyjack: data-center construction is the named growth lever for 2026 and beyond [L-10].
- Agriculture: organic growth + precision-ag adjacency, anchored by Bourgault.
- Capital allocation: 2025 investing outflow $837M ($404M capex); $700M term credit secured early 2024; Q1 2026 FCF $220M.
- Robotics + AI: revenue-generating ambition — contract manufacturer for cobots in North America; humanoid build partnerships ×2; AI brain software collaborations.
03The plant walk
You enter through the security gate. The air is thick with the hum of ventilation, the sharp hiss of compressed air, and the faint metallic tang of cutting fluids. This is a Tier-1 — a company that supplies components directly to the vehicle OEM (auto brand). Tier-2 sells to Tier-1, Tier-3 sells to Tier-2. automotive plant. We will walk eleven stations from receiving dock to shipping. At each station: what you see, who is standing there, what controls the machine, what could fail, and the jargon that gets defined as it appears.
01Receiving dock and raw material staging
SeeForklifts moving stacks of incoming castings, forgings, aluminum billets, machined Tier-2 components. Pallets, returnable totes, labels.
HearBackup beepers, the clatter of wood pallets, dock-door rollups.
SmellDiesel exhaust from forklifts, cold metal, packing oil.
WhoMaterial handlers scanning barcodes; receiving clerks reconciling against the BOM — Bill of Materials, the comprehensive list of parts that go into building one finished unit.; supplier-delivery drivers.
ControlsHandheld barcode scanners feeding the ERP system (SAP, Oracle, or similar).
FailureMislabeling. A batch of billets scanned wrong corrupts traceability downstream. Wrong parts arrive at the line.
02Raw material storage and line-side kitting
SeeRacks of parts broken down into smaller totes (KLT bins — small reusable plastic containers) for line-side delivery. The start of WIP — Work-in-Progress, inventory inside the manufacturing process but not yet finished good..
ControlsKanban — visual scheduling system (cards or digital signals) that triggers replenishment of materials when stock drops below a threshold. cards or digital screens dictate what gets picked next.
FailureKitting errors. Operator puts the wrong casting in a tote → the downstream robot crashes when it grips it.
03Casting and forging
SeeMassive heat-radiating machines. Linamar runs both ductile iron (post-GF Leipzig acquisition) and aluminum (post-Aludyne) processes. Vertical green-sand molding from DISA. Aluminum/magnesium die-casting cells from Bühler. Shot-blasting machines from Sinto cleaning the flash off cooled parts. Robotic deflash and grind cells.
HearDeafening thuds of molds closing, roar of furnaces, abrasive hiss of shot blasting.
SmellBurning sand, ozone, molten metal.
WhoOperators in aluminized heat suits; maintenance techs; the plant GM doing a gemba walk — Japanese for "the real place"; walking the floor where value is created to observe processes directly, rather than reading dashboards from an office..
ControlsSiemens SIMATIC S7-1500 PLCs over PROFINET control furnace temperatures and mold pressures. Data flows via OPC UA to the SCADA system.
RoboticsHeavy-duty heat-shielded robots (often FANUC or KUKA) extract red-hot castings from dies.
FailureSensor degradation. Heat and dust blind optical sensors, raising scrap rate — percent of parts that come off the line defective and unsalvageable. and lowering first-pass yield — percent of parts that pass quality inspection on the first attempt without rework..
Sub-Segment E pivot: Both Linamar acquisitions (GF Leipzig, Aludyne) are foundry/casting operations. If you're walking either, IP67-rated robotics and air-purged camera enclosures are mandatory. Antioch's value is validating sensor performance against oil mist, sand contamination, and glare in simulation before commissioning.
04Machining lines
SeeRows of enclosed CNC (Computer Numerical Control) machining centers and massive transfer lines. GROB customizable machining lines; HELLER TRS modular transfer systems for high-volume parts; Hydromat rotary transfer machines. Robotic load/unload cells at the front of each spindle.
HearHigh-pitched whine of spindle motors, splash of coolant.
SmellSweet, synthetic cutting oil.
WhoCNC set-up techs adjusting tool offsets; cell operators monitoring screens.
ControlsRockwell ControlLogix 5580 controllers managing the cell. Robot logic written in ABB RAPID or standard IEC 61131-3 languages (ladder logic, structured text, function block).
RoboticsABB or FANUC robots tending machines, picking raw castings from bins, loading hydraulic fixtures. This is where the bin-pick-and-place pilot lives. Integrators like Productivity Automation Group build these cells.
FailureChip entanglement. Metal shavings wrap around the part, preventing proper seating in the fixture, causing a crash.
Jargon planted here: The line is designed around a takt time — the required production pace to meet customer demand. Available time / customer demand. If customer needs 60 engines/hour, takt time is 60 seconds.. The robot's cycle time — actual time to complete one operation. Must be shorter than takt time, or you can't keep up. must be faster than takt time. The beat time — actual pace the line is running at right now, which may differ from designed takt. is what the line is doing today. If the robot faults, it impacts OEE — Overall Equipment Effectiveness: Availability × Performance × Quality. World-class is 85%..
05Heat treat
SeeGlowing red parts moving through induction coils or sealed carburizing furnaces. Quench tanks of oil or polymer.
HearLow hum of high-frequency induction generators, sudden splash as parts drop into quench.
SmellBurning oil, ozone, fuel gas (in fired furnaces).
ControlsStandalone PLCs tracking time-at-temperature recipes (recipe management is critical and audited).
FailureQuench distortion. Uneven cooling warps parts and ruins downstream machining tolerances.
Jargon planted here: Heat treat is a critical node in the PFMEA — Process Failure Mode and Effects Analysis. A structured spreadsheet predicting how each step of the process can fail and what controls prevent it.. Parameters are governed by the control plan — the master document for each part-number, listing every critical dimension, every check, how often, by what gauge, what to do if out-of-spec..
06Finishing, grinding, honing
SeePrecision machines removing microns of material. Nagel honing machines, Studer S33 universal cylindrical grinders, JUNKER JUCRANK non-cylindrical grinders.
HearSmooth, high-pitched grinding hiss.
SmellFine mist of grinding coolant.
ControlsIn-line gauges feed dimensional data back to the CNC to automatically adjust tool-wear offsets.
FailureThermal expansion. If coolant temperature fluctuates the part expands and the machine grinds it undersize.
Jargon planted here: Finishing determines Cp, Cpk, Pp, Ppk — process capability indices. Statistical measures of how well a process stays within its tolerance band. Cpk > 1.33 is required by most automotive customers; > 1.67 for safety-critical features.. Operators use SPC — Statistical Process Control. Using control charts (X-bar, R chart) to detect when a process is drifting before it produces out-of-spec parts. and control charts to catch wheel-drift early.
07Sub-assembly cells
SeeOperators and collaborative robots pressing bearings into housings, assembling gear sets, building hydraulic units.
HearSharp thwack of pneumatic presses, the whine of small servo motors.
ControlsPress-fit monitors capturing force-over-distance curves.
FailureMissing components — an O-ring left out, a thrust washer flipped.
Jargon planted here: To prevent missing-component failures the cell uses poka-yoke — Japanese for "mistake-proofing"; a fixture or sensor that physically prevents the wrong action. Example: a sensor that won't let the press cycle unless the O-ring is detected..
08Final assembly
SeeTransmissions or drivelines moving down a Bosch Rexroth TS 2plus transfer system (handles up to 240 kg payloads). DC electric tightening tools from Desoutter or Atlas Copco suspended from balancers.
HearElectronic whir-click of torque guns reaching target. Click-clack of pneumatic clamps.
WhoLine leads monitoring the andon — visual display (often a stack light or overhead board) showing line status. Green = running. Yellow = supplied/help needed. Red = stopped. board. Quality engineers spot-auditing.
ControlsTorque controllers send final torque and angle data to the MES (Manufacturing Execution System) to marry with the part's serial number for traceability.
FailureCross-threading a critical bolt. Missed fastener.
Jargon planted here: Engineers here obsess about line balancing — leveling workload across all stations so no station is starved or blocked. Imbalance creates WIP buildup.. They practice jidoka — "automation with a human touch"; the ability to stop the line automatically the instant a defect is detected, rather than passing the defect downstream. and heijunka — production leveling; smoothing out the mix and volume of products so the line runs at a steady cadence..
09Inspection and metrology
SeeA climate-controlled room off the main floor housing CMMs (Coordinate Measuring Machines) from Zeiss or Hexagon. Optical comparators, surface-roughness gauges.
HearNear silence broken only by the faint beep of a ruby-tipped probe touching metal.
Controls3D point clouds compared against CAD models. Reports auto-uploaded to the quality system.
FailureThe measurement system itself is flawed — the gauge has more variation than the parts you're trying to measure.
Jargon planted here: Before trusting any inspection result, the quality team runs a gauge R&R — Gauge Repeatability and Reproducibility. A statistical study of how much of the observed variation comes from the measurement system vs. the part. If gauge variation is > 30% of total, you can't trust the data..
10Final test
SeeFully assembled units hooked to fluid lines and spinning on test stands. Dürr end-of-line (EOL) testing systems. AVL production-testing rigs. ATEQ or Cincinnati Test Systems leak testers. Integrators like KIE Solution and Burke Porter build these cells.
HearEngines revving (hot test), pressurized air hissing (leak test).
ControlsHigh-speed DAQ (Data Acquisition) systems capturing NVH (Noise, Vibration, Harshness) signatures.
FailureA micro-porosity in an upstream casting causes a leak failure here. The root cause is 8 stations upstream.
Jargon planted here: Passing this test consistently is what enables PPAP — Production Part Approval Process. The standardized 5-level evidence package an automotive supplier submits to the OEM to prove the production process can consistently make the part to spec. The customer signs off; without that signature, you can't ship.. PPAP is the culmination of APQP — Advanced Product Quality Planning. The 5-phase project management framework for launching a new automotive part: plan → design → process design → product/process validation → feedback/improvement., all under IATF 16949 — the global automotive quality management system standard, built on top of ISO 9001 but with much stricter automotive-specific requirements..
11Packaging and shipping
SeeFinished parts loaded into returnable dunnage (custom plastic trays) or customer-specific containers. Final scan, weight check, manifest print.
FailureShipping the wrong part to the OEM. Triggers a chargeback that contributes heavily to COPQ — Cost of Poor Quality. Scrap + rework + warranty + chargebacks + sorting + premium freight. The hidden cost of quality failures..
04Robotics deployment reality
In a production plant the robot arm is the commodity. The product is the integration, safety, and state-machine logic. The 2025 standards reset (ANSI/A3 R15.06-2025 adopting ISO 10218:2025) now embeds cybersecurity in functional safety. The 6-to-18-month deployment tail goes to MES handshakes over ISA-95 and OPC UA, not to motion planning [D-1] [D-2] [D-3].
The 2025 safety standards stack
| Standard | Scope | What an auditor checks |
|---|---|---|
| ANSI/A3 R15.06-2025 (= ISO 10218-1/2:2025) | Industrial robot system safety | Explicit functional safety; cybersecurity embedded in design and deployment [D-1] |
| ISO 13850 | Emergency stop | E-stop device design + placement [D-10] |
| ISO 13855 | Safety distances | Calculated approach distances (K = 2000 mm/s for a hand-arm at human walking speed) [D-9] |
| ISO 14119:2024 | Interlocks | Defeat minimization (anti-tamper, coded interlocks) [D-17] [D-18] |
| ISO/TS 15066 | Cobots | Biomechanical contact limits (force, pressure, energy per body region) [D-21] [D-22] |
| IEC 61496 | Light curtains | Type 4 certification, mounting per ISO 13855 |
What changed in 2025: functional safety requirements that used to be implied are now explicit. Cybersecurity is no longer "IT's problem" — it is part of robot system design. Safety PLCs, networked sensors, and robot controllers are now cyber-physical systems requiring a threat model. If you're designing the bin-pick cell, that means a DFMEA entry for cyber threats alongside mechanical and electrical ones.
Safety device hardware (what you'll see on the cell)
| Device | Examples | Notes |
|---|---|---|
| Safety scanner | Keyence SZ-V, Omron OS32C (104.5mm / 1.3kg), SICK S300 | Flexible zones; sensitive to reflectivity; reduction-of-resolution and stand-still field tricks for in-line setups |
| Light curtain (Type 4) | SICK / Keyence / Omron / Banner | Per IEC 61496; mount per ISO 13855 distances |
| Interlocks | Schmersal, Pilz, Euchner — ISO 14119 anti-tamper devices | Coded interlocks prevent override-by-zip-tie |
The integrator handoff
| Party | What they own |
|---|---|
| Robot OEM (ABB, FANUC, KUKA, Yaskawa) | Arm, controller, base software, extended warranties (e.g., KUKA WarrantyPro = 5 yr + annual PM [D-25]; ABB and FANUC offer similar tiers [D-26] [D-27]) |
| System integrator (Bastian, JR Automation, Genesis Systems, Acieta, Productivity Automation, ATC, ACE Industrial) | Cell design, EOAT integration, vision integration, path programming, commissioning, operator training. Specialty matters: Genesis = welding (IPG-owned); Bastian = warehouse (Toyota-owned); Acieta = FANUC platinum; JR Automation = Hitachi broad portfolio. |
| End-user (Linamar) | Site prep, utilities, operator training, ongoing ops + maintenance |
ROI math and what kills it
Tier-1 ROI targets are 18-36 month paybacks on capex. Typical inputs: labor displaced (1-2 operators per cell, $50-90K/yr loaded), throughput uplift, scrap reduction. Capex: arm + EOAT + vision + integration + safety + fixtures + commissioning runs $250K to $1M+ per cell. The IFR World Robotics 2024 reports industrial-robot installations continue to exceed 500,000 units annually globally despite macro headwinds [D-29].
What kills the ROI:
- Scope creep on MES integration (the integrator quoted point-to-point; the customer wants full plant-wide visibility)
- Uptime degradation (nuisance trips, network jitter) — a 2-3 point drop in uptime dominates the calculation
- Operator workarounds (interlocks bypassed → safety incidents + downtime)
Production failure modes
| Failure | Symptom | Root cause | Prevention |
|---|---|---|---|
| Operator workarounds | Interlocks bypassed | Keys left in machines / shortcut culture | ISO 14119 anti-tamper, trapped-key systems, audits |
| Network jitter | Missed sync windows | No time synchronization | CIP Sync (EtherNet/IP) or PTP [D-32] |
| EtherNet/IP drops | Connection flaps | Secondary port misconfig | Network QoS, validated reference architectures |
| Poor part presentation | Vision finds nothing or wrong thing | Bin variability, occlusion, glare | Mechanical part presentation, lighting design, simulate before deploy |
| MES handshake gaps | Robot waits for signal that never comes | OPC UA state machine bug, recipe mismatch | Rigorous state-machine testing in simulation pre-commissioning |
| EOAT wear mid-cycle | Gripper drops part | Suction-cup fatigue, jaw wear | Spares kit on-site, scheduled EOAT replacement |
05Bin-picking specifically
This is Linamar's named pilot use case. Dana on the call: "picking a part from a bin... orienting that part to load a machine... limited mobility... two to four machines." The hard part is not the grasp. The hard part is the long tail: oily castings, glare, occlusion, the integration tail to MES, and a mispick rate budget of <0.1% in automotive (one wrong part to a CNC fixture can break a $50-200K tool).
Gripper choice trade-offs
| Gripper | Pros | Cons | Best for |
|---|---|---|---|
| Vacuum | Simple, fast | Fails on porous, oily, or irregular surfaces | Flat clean items, sheet metal |
| Parallel jaw | Universal; predictable | Slow to retool for new part shapes | Castings, machined parts |
| Magnetic | Strong, simple | Ferrous only; residual magnetism issues | Forgings, steel stampings |
| Soft gripper | Low force, deformable parts | Slow, fragile | Specialty / fragile |
| Quick-change EOAT | Multi-SKU flexibility | Mid-cycle swap time + reliability concern | High-mix lines |
Vision vendor faceoff
| Vendor | Tech | Sweet spot | Notable claim |
|---|---|---|---|
| Photoneo (Slovak) | PhoXi 3D scanner — structured light | High-res static scenes | Scanners ~€10K; Bin Picking Studio software |
| Mech-Mind (Chinese) | Mech-Eye 3D cameras + Mech-Vision software | Broad bin-picking, cartons, metal parts | Cited 6-12s pick cycles at >99.5% success on production cases |
| Zivid (Norwegian) | 3D color + depth | Smarter factories, AI/ML projects | Premium fidelity |
| Apera AI (Canadian) | 4D Vision — AI-driven, uses off-the-shelf 2D cameras | Software-led | Cycle as fast as 0.3s (3 Hz); claims >99.99% reliability via simulation training |
| Cognex (US) | In-Sight + 3D-A1000 / A5000 | Dominant in factory inspection; expanding to robot vision | Industrial-grade reliability |
| Keyence (Japanese) | CV-X / RB-series | Integrated vision system | Premium service + support |
| Pickit (Belgian) | Vision-as-a-service for integrators | Easy plug-in for system integrators | Mid-market |
Grasp planning in 2026
| Regime | Examples | Where it wins |
|---|---|---|
| Analytical (force-closure, friction cone) | Classical robotics | Strong guarantees when you have accurate geometry — gears, machined parts |
| Data-driven (synthetic training) | Dex-Net 2.0 / 4.0 (Berkeley, Mahler et al.) | Pre-trained policies on millions of simulated grasps with domain randomization — solid baseline for known SKU sets |
| RL (vision-based) | QT-Opt (Kalashnikov et al.) | Scalable deep RL — research-leading on dynamic manipulation |
| Learning-based VLA + diffusion | Pi-0 (Physical Intelligence), RT-1/RT-2, OpenVLA, Octo, RDT, Helix (Figure), Diffusion Policy (TRI/Columbia) | Broad generalization; harder to certify failure modes; production deployments still early |
Production-readiness KPIs (Run@Rate / SAT criteria)
- Pick success ≥ 99.5% sustained across an 8-hour drift test
- Mispick rate ≤ 0.1% (1 in 1,000 — anything worse breaks tooling in automotive)
- Cycle time within takt budget (typically 6-15 cycles/min for a 10-12 kg payload arm)
- MTTR ≤ 60 seconds (time to clear and reset after a failure)
- Empty-bin detection accuracy 99%+ (false-empty wastes cycles; false-positive causes crash)
- Throughput maintenance over the shift (no degradation from EOAT wear, lighting drift, temperature)
The integration tail (where the 6-18 months really go)
- ERP triggers (parts call from MRP) → how the cell knows what to pick next
- MES handshake over OPC UA: Cell-Ready → Part-Located → Part-Loaded → Machine-Busy → Machine-Ready → Part-Removed
- Downstream machine load — CNC fixture alignment tolerance, transfer-line synchronization
- Scrap routing — where rejected parts go and how they're traced
- Error reporting — alarm propagation up to the andon, SCADA, plant-wide visibility
- Recipe management — when the same cell runs multiple part-numbers, the changeover
Where Antioch slots in
Antioch's deployment-trust layer. Physical iteration cannot test the cross-product of vision system × gripper × grasp-planner × part-population × lighting × bin-fill at the speeds AI development now demands. Simulation can. The pitch is: validate the candidate combination in sim, calibrate against live data via pair-testing, surface the failure modes the demo never showed, certify Run@Rate before commissioning. Apera AI publicly trains on 1M+ simulated cycles; Berkeley's Dex-Net is built on synthetic data. Antioch is the same primitive applied to the customer's actual cell, with calibration to the customer's actual telemetry — not a generic asset library.
06Tier-1 analogues — what the peers are publicly doing
Twelve Tier-1 industrial manufacturers Linamar will measure itself against. Three of them have explicitly named humanoid programs (Magna, Bosch, Schaeffler). Several have NVIDIA Omniverse / OpenUSD digital-twin programs. The pattern: Tier-1s are positioning as both users of and component suppliers to humanoid OEMs. Linamar's "we want to be a North American component manufacturer for someone" is now segment-standard.
CompanyFitPublic AI/robotics signal
Magna International
Aurora, Ontario · $42.8B FY24 High Strategic partnership + equity in Sanctuary AI for "general purpose AI robots for deployment in Magna's manufacturing operations" (Phoenix humanoid + Carbon AI). Includes "multi-disciplinary assessment of improving cost and scalability of robots using Magna's automotive product portfolio." Scaling operational AI on NVIDIA Omniverse + Cosmos with VLM exploration. [A-3] [A-5]
Aurora, Ontario · $42.8B FY24 High Strategic partnership + equity in Sanctuary AI for "general purpose AI robots for deployment in Magna's manufacturing operations" (Phoenix humanoid + Carbon AI). Includes "multi-disciplinary assessment of improving cost and scalability of robots using Magna's automotive product portfolio." Scaling operational AI on NVIDIA Omniverse + Cosmos with VLM exploration. [A-3] [A-5]
Continental AG
Hanover · €40B High March 2025: deployed 7 AMRs at ContiLifeCycle Hanover-Stöcken retreading plant (sensors + 360° cameras + AI nav). Building OpenUSD virtual factories with SoftServe + NVIDIA Omniverse. [A-4] [A-13]
Hanover · €40B High March 2025: deployed 7 AMRs at ContiLifeCycle Hanover-Stöcken retreading plant (sensors + 360° cameras + AI nav). Building OpenUSD virtual factories with SoftServe + NVIDIA Omniverse. [A-4] [A-13]
Bosch (incl. Rexroth)
Stuttgart High Major strategic partnership with NEURA Robotics to "drive the industrial scaling of humanoid robotics and Physical AI." Explicit: "potential supply of robotic components by Bosch, as well as possible final assembly and motor production for humanoid robots." Bosch Rexroth Digital Product Twin + Microsoft Manufacturing Co-Intelligence. [A-1] [A-11]
Stuttgart High Major strategic partnership with NEURA Robotics to "drive the industrial scaling of humanoid robotics and Physical AI." Explicit: "potential supply of robotic components by Bosch, as well as possible final assembly and motor production for humanoid robots." Bosch Rexroth Digital Product Twin + Microsoft Manufacturing Co-Intelligence. [A-1] [A-11]
Schaeffler
Herzogenaurach · €25B+ (post-Vitesco) Med Future-oriented partnership with NEURA Robotics including offtake agreement on humanoid components. [A-2]
Herzogenaurach · €25B+ (post-Vitesco) Med Future-oriented partnership with NEURA Robotics including offtake agreement on humanoid components. [A-2]
Denso
Kariya · JPY 7.1T Med DENSO WAVE robotics subsidiary (manufactures and sells industrial robot arms). IREX 2025 product showcase. 2025 Integrated Report commits to digital transformation; no Omniverse / DELMIA specifically. [A-7] [A-9]
Kariya · JPY 7.1T Med DENSO WAVE robotics subsidiary (manufactures and sells industrial robot arms). IREX 2025 product showcase. 2025 Integrated Report commits to digital transformation; no Omniverse / DELMIA specifically. [A-7] [A-9]
ZF Friedrichshafen
Friedrichshafen · €42B Med Digital Manufacturing Platform (DMP) — "By 2026, all plants are to be connected to the DMP... take a leading position in the international arena when it comes to smart production." No explicit humanoid program disclosed. [A-15]
Friedrichshafen · €42B Med Digital Manufacturing Platform (DMP) — "By 2026, all plants are to be connected to the DMP... take a leading position in the international arena when it comes to smart production." No explicit humanoid program disclosed. [A-15]
John Deere
Moline, IL · $60B Med Jan 2025 Gen-2 autonomy kit (advanced CV + AI + camera). Autonomy Precision Upgrade lowers barrier to entry for autonomous farming. John Deere Operations Center as digital solutions platform. Cultural fit is high but use case is outdoor/unstructured agricultural. [A-19] [A-20]
Moline, IL · $60B Med Jan 2025 Gen-2 autonomy kit (advanced CV + AI + camera). Autonomy Precision Upgrade lowers barrier to entry for autonomous farming. John Deere Operations Center as digital solutions platform. Cultural fit is high but use case is outdoor/unstructured agricultural. [A-19] [A-20]
Emerson
St. Louis, MO · $17B Med DeltaV Mimic Digital Twin — documented case of plant starting production 6 weeks ahead of schedule, ROI 8x over. AspenTech Subsurface Intelligence (ASI). Primarily process automation (fluids/chemicals), not discrete manufacturing. [A-26]
St. Louis, MO · $17B Med DeltaV Mimic Digital Twin — documented case of plant starting production 6 weeks ahead of schedule, ROI 8x over. AspenTech Subsurface Intelligence (ASI). Primarily process automation (fluids/chemicals), not discrete manufacturing. [A-26]
Aisin
Kariya · JPY 4.4T Low PLM transformation; GenAI for audit records and data analysis. No advanced robotics or Omniverse-class digital twin publicly disclosed. [A-16]
Kariya · JPY 4.4T Low PLM transformation; GenAI for audit records and data analysis. No advanced robotics or Omniverse-class digital twin publicly disclosed. [A-16]
ITW
Glenview, IL · $16B Low ~84 decentralized business units (the Linamar plant-GM-autonomy comp). No unified corporate digital-twin or robotics strategy. Selling unit-by-unit is inefficient. [A-22]
Glenview, IL · $16B Low ~84 decentralized business units (the Linamar plant-GM-autonomy comp). No unified corporate digital-twin or robotics strategy. Selling unit-by-unit is inefficient. [A-22]
Honeywell
Charlotte, NC · $36B Low Q3 FY2025 earnings: "Increasingly, customers across end markets face similar structural challenges such as skilled labor shortages, aging infrastructure, operational inefficiencies..." But agreed to sell Intelligrated + Transnorm (warehouse automation) to American Industrial Partners — actively exiting the segment Antioch simulates. [A-24] [A-25]
Charlotte, NC · $36B Low Q3 FY2025 earnings: "Increasingly, customers across end markets face similar structural challenges such as skilled labor shortages, aging infrastructure, operational inefficiencies..." But agreed to sell Intelligrated + Transnorm (warehouse automation) to American Industrial Partners — actively exiting the segment Antioch simulates. [A-24] [A-25]
Hitachi Astemo
Tokyo Low In ownership transition: Honda acquiring additional 21% equity (Dec 2025). Net-new experimental software capex likely frozen until restructuring settles. [A-17]
Tokyo Low In ownership transition: Honda acquiring additional 21% equity (Dec 2025). Net-new experimental software capex likely frozen until restructuring settles. [A-17]
Top 3 most aligned
- Magna — Sanctuary AI humanoids + Omniverse digital twins. Exact stack, exact problem.
- Continental — OpenUSD virtual factories + AMR fleet deployment.
- Bosch — NEURA humanoid scaling + Digital Product Twin platform.
The 3-sentence answer if Linamar asks "who else is doing what you do"
Your most aggressive peers are no longer just buying robots; they are using platforms like NVIDIA Omniverse to build digital twins that simulate and validate AI-driven humanoids and AMRs before they ever hit the factory floor, as seen with Magna's Sanctuary AI pilot and Continental's OpenUSD virtual factories. Tier-1s like Bosch and Schaeffler are actively partnering with humanoid OEMs like NEURA Robotics not just to use the robots, but to become the primary suppliers of their internal components and motors. Antioch provides the exact simulation and validation layer required to prove these robotic deployments work in a digital twin first, so you don't lose millions on stalled physical pilots.
07Jargon glossary
Quick reference. Every term here is also defined inline where it first appears in the plant walk. Skim, then come back when you need to look one up fast.
Operations rhythm
Takt time
Required pace to meet customer demand. Available time / customer demand.If customer wants 60 engines/hour → takt = 60s.
Cycle time
Actual time to complete one operation. Must be < takt time.The ABB cell loads a casting in 45s. Takt is 60s. Margin: 15s.
Beat time
Pace the line is actually running at right now.Designed for 60s takt; running at 72s beat = line is behind plan.
OEE
Overall Equipment Effectiveness = Availability × Performance × Quality. World-class: 85%.Cell ran 90% of shift, at 95% of design speed, with 99% good parts → OEE 84.6%.
MTBF / MTTR
Mean Time Between Failure / Mean Time To Repair.MTBF up = reliability up; MTTR down = downtime per failure down.
Bottleneck
Slowest station; dictates max throughput.The welding cell at 30 parts/hr caps the whole line.
Throughput
Total volume passing through the system per unit time.Line balancing
Distributing work evenly across stations so none waits.Takt-paced vs asynchronous
Parts move on a strict timer (takt) vs whenever the next station is ready (async/buffered).WIP
Work-In-Progress. Inventory inside the process.Quality systems
PPAP
Production Part Approval Process. 5-level evidence package showing the supplier can consistently produce the part. Customer signs → you can ship.APQP
Advanced Product Quality Planning. 5-phase framework for launching a new automotive part.IATF 16949
Global automotive QMS standard. Built on ISO 9001 + strict automotive-specific requirements.ISO 9001
Baseline quality management system standard.FMEA / PFMEA / DFMEA
Failure Mode and Effects Analysis. Process / Design. The spreadsheet predicting failures and controls.Control plan
The master document listing every critical dimension, check, frequency, gauge, and out-of-spec response per part.Cp, Cpk, Pp, Ppk
Process capability indices. Cpk > 1.33 typical automotive requirement; > 1.67 safety-critical.Gauge R&R
Repeatability & Reproducibility study. How much of observed variation comes from the measurement system. Must be < 30%.SPC / Control charts
Statistical Process Control. X-bar and R charts detect drift before defects.First-pass yield
% of parts passing inspection first try without rework.Scrap rate
% of parts thrown away as unsalvageable.Rework
Fixing a defective part to make it good (vs scrap).COPQ
Cost of Poor Quality. Scrap + rework + warranty + chargebacks + premium freight.Run-at-rate / PTR
Production Trial Run. Run the line at full design speed to prove it works pre-SOP (Start of Production).Lean / Toyota Production System vocabulary
Kaizen
Continuous, incremental improvement.Kanban
Visual replenishment signal.Andon
Visual / audio alarm board showing line status.Poka-yoke
Mistake-proofing (a fixture that only accepts a part in the correct orientation).5S
Sort, Set in order, Shine, Standardize, Sustain.Gemba / gemba walk
"The real place"; going to the floor to observe processes directly.Jidoka
Automation with a human touch. Auto-stop on defect detection.Heijunka
Production leveling — smoothing volume + mix.5 Whys / Fishbone
Root-cause analysis tools (Ishikawa diagram).Pareto chart
Bar chart of defect causes; 80% of problems from 20% of causes.Engineering / business basics
Tier-1 / Tier-2 / Tier-3
Supplier hierarchy. T1 → OEM. T2 → T1. T3 → T2.BOM
Bill of Materials. The recipe of parts.COGS
Cost of Goods Sold.DFM / DFA
Design for Manufacturability / Assembly.SOP
Start of Production (the customer launch milestone).EOL test
End-of-Line test (cold, hot, leak).CMM
Coordinate Measuring Machine.EOAT
End of Arm Tooling (gripper, sensor, force-torque).08Fluency phrases — what they say, what it means, how you respond
Twenty sentences a Director-level robotics lead or plant-facing PM will say on the walk. The annotation tells you what they actually mean and what an Antioch-fluent response looks like.
We're not happy with the OEE on this cell.
The cell is underperforming on Availability × Performance × Quality.
Which factor is dragging — are you seeing micro-stops (performance) or hard faults (availability)?
The PPAP timing is going to be tight on this program.
Worried about hitting the customer-approval gate before SOP. Ramp risk.
Let's review the PFMEA. Can we simulate the high-risk stations to pull validation forward?
Have you done a gauge R&R on that?
Is the measurement system itself the source of variation?
Right — we need to isolate tool wear from sensor drift before we train the AI on this data.
What's the bottleneck station dictating our throughput?
Where is the constraint that limits total output?
If it's CNC load/unload, we can look at optimizing the RAPID path; in sim first.
Are we running takt-paced or asynchronous here?
Do parts move on a strict timer or buffer between stations?
If it's asynchronous, we have buffer time to run edge-AI inference without starving the next op.
The Cpk on this bore is drifting.
The machining process is slowly moving out of tolerance.
Let's tie spindle load data to the SPC chart to predict tool wear before parts cut undersize.
We need to update the Control Plan if we add this vision system.
IATF 16949 requires documentation of any new quality check.
Agreed. We'll map AI confidence scores to the existing pass/fail criteria for the auditor.
Is this a hard PL/d safety interlock or just a software stop?
Does this rely on a physical safety relay or just PLC code?
We will never override a SIL/PL-rated safety circuit with an AI agent.
What's the MTTR when the robot drops a part?
How long to clear the crash and reset the cell?
If MTTR is high, we should implement an auto-recovery routine in the PLC.
We're seeing too much WIP building up before heat treat.
Line unbalanced; cash tied up in unfinished inventory.
Can we use predictive scheduling to smooth the heijunka flow into the furnaces?
Did this fail at the EOL hot test or the leak test?
Trying to isolate defect origin — assembly or upstream casting.
If it's a leak failure on the ATEQ rig, look at upstream die-casting parameters first.
We need a poka-yoke for this kitting process.
Operators keep putting the wrong parts in the bin.
A simple overhead camera with a classification model can verify bin contents before it leaves staging.
The scrap rate is killing our COPQ.
Throwing away too much money on bad parts.
Let's build a Pareto chart from MES defect codes and target the top offender.
Is the data coming over PROFINET or OPC UA?
How easily can I extract tags from the PLC?
We prefer OPC UA for the edge gateway so we don't disrupt deterministic PROFINET I/O.
Let's do a gemba walk to see the chip evacuation issue.
Look at the physical machine, not the dashboard.
Yes — I want to see if the robot's EOAT is fouling on coolant overspray.
We need to hit run-at-rate by Q3.
Line must prove full production speed for the customer signoff.
We'll deploy in shadow mode during the PTR so we don't impact cycle time during validation.
The first-pass yield on the transfer line is dropping.
Parts are failing inspection and being routed to rework.
Let's correlate HELLER spindle vibration with downstream CMM rejects.
Can we pull the torque curves from the Desoutter controllers?
I want the rich fastening data, not just the pass/fail bit.
Yes — rundown-angle analysis can predict cross-threading before the final torque is reached.
This design has terrible DFM.
Part is incredibly hard to manufacture.
If we can't change the casting, we need adaptive robot paths to handle the dimensional variation.
Let's 5-Why this downtime event.
Find root cause, not symptom.
Agreed — sensor failure, PLC logic trap, or mechanical bind?
09Per-attendee cheat sheet
Frame the room. Dana decides commercially. Lina gates plant access. Mackenzie sets the technical evaluator bar. Tom keeps the relationship moving. Anchor every response in the deployment-gap framing: 95% in the lab vs. 99.5% on a line where one mispick breaks a $200K tool.
Public-background read added 2026-05-15. Dana is not just "the robotics director"; she chairs Linamar's AI Council and is visibly attached to the Microsoft Copilot rollout. Lina is not just a PM; she is a P.Eng LEAP operator with plant launch, PFMEA/APQP, e-axle, waste-elimination, and global cost-savings reps. Tom is a founder/commercialization operator from New Zealand. Mackenzie has the thinnest public trail, so treat her May 7 technical questions as the source of truth.
Dana Sharp
Director, Linamar Robotics · Economic Buyer · AI Council Chair since 2022 · finance / process automation / data / app-dev arc
Focus: Risk, capex, governance, ROI, commercial framing. Public profile shows Director - Linamar Robotics from Mar 2025 and AI Council Member - Chair from Jun 2022. SafetyMag also places her with Linda Hasenfratz and Jim Jarrell at Microsoft HQ for Linamar's Copilot / AI ergonomics rollout. [P-1] [P-4]
If she says: "I can't risk an AI hallucination crashing a $2M GROB machine."
We don't deploy directly to the PLC. We use simulation pair-testing: run the AI behavior in a calibrated digital twin, score it against production constraints, and only then discuss what ever gets near a Rockwell / Siemens control layer.
What earns her trust: connect Antioch to Linamar's broader AI rollout, not just robotics. Say: "The same pattern as Ergo Assist applies here: use AI where it removes low-value friction, keep technical hazard decisions auditable." She will respect governance language.
Lina Qamar
P.Eng · Program Manager, Manufacturing & Engineering Robotics · LEAP graduate · plant-ops champion
Focus: APQP / PPAP timing, run-at-rate, OEE, plant-GM relationships, daily ops. Public profile says she led a Waste Elimination rotation across 70+ global facilities with $52M CAD in savings, worked APQP / PFMEA / control plans / PHSR / ESA audits, and was lead project engineer on e-axle components in the Gear Lab. [P-2]
If she says: "Any changes to the cell will require a new PPAP and update to the Control Plan."
Exactly. We should structure this so the validation evidence drops into your existing APQP / PFMEA / Control Plan workflow instead of creating a parallel science project.
What earns her trust: ask concrete plant questions. "Which part family, which takt, which fixture, which plant GM, what failure mode would kill this?" She has been close to line launches and safety paperwork. Generic robotics excitement will bounce off her.
Mackenzie Kuntz
Data Science + AI Specialist · AI-robotics technical evaluator · public profile unresolved
Focus: Data pipelines, tag mapping, model drift, VLA + RL stack choices. Lightweight public search did not resolve a credible Linamar profile. Do not overfit biography. Use her May 7 call evidence: pre-data-collection, software-to-physical-AI transition, cautious about naming vendors, and interested in how Antioch fits a phased data collection / fine-tuning loop.
If they say: "Our Siemens S7-1500s are locked down. How are you getting the data?"
We don't need to disturb the control network to prove value. Start with the telemetry you already expose through SCADA / historian / robot logs, then map the missing tags only after the pilot failure modes are explicit.
What earns her trust: make the engagement mechanics legible: inputs, logging, eval harness, data schema, how pair-testing compares model versions, and what happens when the real cell disagrees with sim. Keep Physical Intelligence / Pi-0 references as hooks, not a lecture.
Tom Schuyt
Commercialization lead · ex-founder from NZ · Greengrower co-founder/CEO 2020-2024 · PwC corporate finance background
Focus: NRE costs, SLAs, value capture, the "manufacture for someone" component-supply angle. Public profile frames him as an ex-founder in food/agtech, "building the future of food by growing more with less," with prior PwC corporate finance experience. That maps cleanly to a founder/operator evaluating whether Linamar can commercialize robotics capability beyond internal deployment. [P-3]
If he says: "Integrators charge a fortune for custom vision. How does this scale?"
The scalable layer is not custom integration at each station. It is a repeatable validation workflow across part families, robot setups, AI stacks, and plant conditions. That matters if Linamar is both deploying robots internally and preparing to manufacture robotics components for OEMs.
What earns his trust: treat him like a commercialization founder, not a generic BD contact. Tie the pilot to proof he can reuse internally with Dana/Lina and externally with future robotics OEM partners.
10Field walk checklist
A focused list of things to ask, observe, and position once your boots are on the floor.
Ask
- Which of the 86 facilities hosts the pilot — Vehcom? Linex? a Leipzig / Aludyne site?
- Who is the plant GM at the pilot site and what is their robotics posture?
- What's the current bin-pick state — manual? partial-auto with a different vendor? on a wishlist?
- Which specific part family — casting, forging, gear, structural? Cycle time target?
- What does the data infrastructure look like today — what telemetry comes off the 5,000 ABB robots, where does it land?
- What's the OPC UA tag-exposure story from the Siemens/Rockwell PLCs at the pilot site?
- Who handles the integrator role — internal Linamar team or a named firm (Acieta, JR Automation)?
- What's the procurement gate above Dana for a $50K pilot? FY26 budget or new line item?
- Naming permission for Linamar as a reference — soft-asked at scope-doc stage, not first call.
- Which OEMs are Linamar targeting for the cobot/humanoid contract-manufacturing motion?
Observe
- Operators leaving keys in machines (defeated interlock = audit risk)
- Andon board state — green/yellow/red distribution at any moment
- Bin variability at the bin-pick candidate cell: depth, fill, glare, oil
- Robot vendor / model labels (most ABB but flagging KUKA, FANUC, Yaskawa pockets)
- Cable-tray density (a proxy for IT/OT segmentation maturity)
- Existing camera / vision systems: Cognex, Keyence, others?
- Whether the plant runs takt-paced or asynchronous (look at conveyor cadence)
- Quality-gate stations — CMM room, EOL test cells, leak testers
Position
- Open with the deployment-gap framing — 95% lab, 99.5% production, sim closes it.
- Cite Jarrell's earnings-call humanoid quote when component-mfg comes up.
- Name peers without naming Linamar: "comparable Tier-1s like Magna are running this on Omniverse with Sanctuary."
- Anchor the pilot in IATF 16949 / PPAP language. The cell change must be doc-able.
- Don't promise full-stack autonomy. Promise validation. Pair-testing earns the right.
- Sequence Phase 1 (single-stack at Vehcom) → Phase 2 (combinatorial across morphologies at Aludyne/Leipzig).
11Follow-up prompts
Each card is a self-contained prompt you can drop into a fresh agent. The copy button puts it on your clipboard.
P1 · Build the per-attendee Granola-cleaned prep blurb for Dana / Lina / Mackenzie / Tom
Write a 250-word per-person prep blurb for the four named Linamar attendees on Tuesday 2026-05-19 (Dana Sharp, Lina Qamar, Mackenzie Kuntz, Tom Schuyt). Pull from the Linamar org page, learnings sidecar, MEDDPICC, the 2026-05-07 meeting transcript, and any public sources (LinkedIn, conference talks). Include for each: career arc, decision style, what to lead with, what to avoid, the one question to ask them directly. Output to agent/prep/2026-05-19-linamar-attendees.md. Use the deployment-gap framing and the Jarrell Q1-2026 robotics quote as touchstones.
P2 · Convert the bin-pick scope doc into a Linamar-shaped Antioch proposal
Linamar is sending a scope doc for the bin-pick-and-place pilot. When it arrives, draft an Antioch response proposal scoped as Phase 1 (single robot / single environment / single stack at Vehcom Manufacturing in Guelph) → Phase 2 (combinatorial validation across morphologies at Aludyne or GF Leipzig). Fixed-fee Phase 1, year-2 platform + GPU usage. Use the deployment-gap framing. Map the validation outputs (Cpk capability data) to slot into Linamar's PPAP submission per IATF 16949. Reference the Antioch Amazon DFM loop as the model. Output to agent/drafts/2026-05-XX-linamar-pilot-proposal.md.
P3 · Surface the actual humanoid-OEM landscape for Linamar's component-mfg ambition
Linamar publicly stated they want to be the North American contract manufacturer for cobots and have two humanoid build partnerships. Research which humanoid OEMs are most likely candidates: Figure AI, 1X, Apptronik, Sanctuary AI, Agility Robotics, Boston Dynamics, Unitree, Sunday Robotics, Mentee, Persona, Mantis, Mythical, NEURA, Foundation, EngineAI. For each: NA component sourcing posture, public partnerships with Tier-1 industrial mfgrs, manufacturing readiness level, fundraise stage. Identify the 3 most likely Linamar OEM partners and the 3 least likely. Output to projects/gtm/linamar-humanoid-oem-landscape.md.
P4 · Deep dive on Vehcom Manufacturing specifically
Research Vehcom Manufacturing at 74 Campbell Rd, Guelph, ON. What products are made there, what end customers, what's the public history (Linamar's nameplate, prior names like Vehcom Manufacturing LP, ownership), what permits/approvals exist (Environmental Compliance Approval), what's been publicly disclosed about automation maturity. Compare to Linex Manufacturing (355 Massey Rd, Guelph) — which is the better bin-pick-and-place pilot site? Pull from Ontario ERO filings, Linamar SEDAR disclosures, LinkedIn employee profiles, local press. Output to agent/research/2026-05-XX-vehcom-deep.md.
P5 · Build the Aludyne + GF Leipzig replication kit narrative
If the Vehcom bin-pick pilot succeeds, Phase 2 expands to Aludyne North America facilities (light-metal castings: knuckles, subframes, control arms, axle housings) and GF Leipzig (large ductile iron castings, HD truck axle program). Write the replication-kit story: which specific Aludyne and Leipzig facilities, what parts at each, what the part-presentation challenges are (oily castings, foundry environment, sand cores), what gets parameterized in the digital twin to support replication, what the financial argument is for multi-plant ROI. Output to projects/gtm/linamar-phase2-replication.md.
P6 · Map Linamar's actual ABB controller and SCADA stack across facilities
Linamar has ~5,000 industrial robots (mostly ABB). Research what's public about their PLC + SCADA + MES stack: which Rockwell ControlLogix versions, which Siemens TIA versions, what SCADA platform (FactoryTalk View / Wonderware / Ignition / WinCC), what MES, what ERP. Look at Linamar IT job postings (LinkedIn, Indeed) for required certifications and stack mentions. Look at integrator case studies that name Linamar. Look at ABB / Rockwell partner page customer references. Output a "Linamar plant-floor IT/OT stack — what we publicly know" map to projects/gtm/linamar-ot-stack.md.
12References
Linamar primary sources
- Dana Sharp public LinkedIn snippet: Director - Linamar Robotics from Mar 2025; AI Council Member - Chair from Jun 2022; prior IT Manager, Global Product Management, Data & Analytics.
- Lina Qamar public LinkedIn snippet: P.Eng; Program Manager, Manufacturing & Engineering Robotics; LEAP rotations; waste elimination across 70+ global facilities; $52M CAD savings; APQP / PFMEA / control-plan work; Gear Lab e-axle project engineering.
- Tom Schuyt public LinkedIn snippet: ex-founder; Greengrower co-founder/CEO 2020-2024; advisor; prior PwC New Zealand corporate finance.
- Canadian Occupational Safety, 2026-03-11: Linamar Ergo Assist / Microsoft Copilot article naming Director of Linamar Robotics Dana Sharp and describing Linamar's internal AI rollout.
- Apera AI, 2024-01-30: Hastech / Linamar random bin-picking case with ABB Robotics and Apera AI across three cells.
- Linamar Q1 2026 Earnings Call (PDF) — linamar.com
- Investing.com — Linamar Q1 2026 earnings call transcript
- Linamar press — Completes Acquisition of GF Leipzig (Dec 2025)
- Linamar press — Acquires Aludyne North American Operations (Nov 2025)
- Die Casting Org — Linamar Ontario Gigacasting plant
- Ontario Environmental Registry — Vehcom Manufacturing Environmental Compliance Approval
- Canada Chamber of Commerce — Vehcom Manufacturing Guelph listing
- Industrial Guide — Linex Manufacturing Inc., Guelph
- PrivCo — Linamar Corporation profile
- Linamar — 2025 Annual Report MD&A
- Georg Fischer — media release on iron foundry divestiture to Linamar
- Linamar press — Aludyne NA acquisition closing
- Linamar press — Dura-Shiloh deal closing (Aug 2023)
- Linamar press — Bourgault deal closing (Feb 2024)
- Linamar Q3 2025 Press Release
Deployment + safety + bin-picking sources
- ANSI/A3 R15.06-2025 — robot safety overview (ANSI blog)
- ISO 10218-1:2025
- ANSI/A3 R15.06-2025 webstore
- ISO 10218-2:2025
- ISA-95 standard
- OPC UA Specification (OPC Foundation)
- Apera AI — Automated Robotic Bin Picking
- Q-Directive — Run @ Rate / Capacity Verification
- Pilz — EN ISO 13855 safeguarding distances
- IDEC — ISO 13850 e-stop
- ISO 13851:2019 — two-hand control
- Keyence SZ-V Safety Scanner
- Direct Target Products — Keyence SZ-V datasheet
- Omron OS32C Safety Laser Scanner
- SICK S300 datasheet
- Industrial Safety Sensor — Type 4 Light Curtain guide
- ISO 14119:2024 — Interlocks
- ISO 14119:2024 — defeat minimization
- ANSI/A3 R15.06-2025 (alt)
- Intertek — Robotics testing & certification
- ISO/TS 15066:2016 — iTeh Standards
- DGUV — Collaborative robot systems
- Unchained Robotics — turnkey commissioning
- SunGene — FAT/SAT acceptance criteria
- KUKA WarrantyPro
- ABB Service Agreements
- FANUC Service Contracts
- IPG Genesis Systems
- IFR World Robotics 2024 Executive Summary
- IFR World Robotics 2024 Press Conference
- OSHA Citation 1581047 (interlocks)
- ODVA — CIP Sync
- Rockwell — Scalable Time Distribution
- Photoneo PhoXi 3D Scanner
- Mech-Mind — Robotic Bin Picking with 3D Vision
- Berkeley Automation — Dex-Net
- Mahler et al. — Dex-Net 2.0 (arXiv 1703.09312)
- Mahler et al. — Ambidextrous Grasping (Science Robotics)
- Kalashnikov et al. — QT-Opt (arXiv 1806.10293)
- SICK Connect — SICK + KUKA bin-picking case study
Analogue sources (Magna / Continental / Bosch / Schaeffler / Denso / ZF / John Deere / Aisin / Hitachi Astemo / ITW / Honeywell / Emerson)
- NEURA Robotics + Bosch partnership announcement
- Schaeffler + NEURA Robotics future-oriented technology partnership
- Sanctuary AI — Magna automotive deployment partnership
- NVIDIA Developer — Continental + SoftServe OpenUSD virtual factories
- Magna — Operational AI with NVIDIA Omniverse + Cosmos
- Magna — 2025 Annual Report
- DENSO Robotics vendor profile
- DENSO WAVE IREX 2025
- DENSO 2025 Integrated Report
- DENSO WAVE Global
- Bosch Rexroth — Digital Product Twin
- Microsoft — AI in Action (Bosch collaboration)
- Automated Warehouse — Continental AMRs at Hanover
- Continental press release — Hanover-Stöcken AMRs (Nov 2025)
- ZF — Digital Manufacturing Platform
- Aisin Integrated Report 2025
- Astemo news release — Honda equity increase (Dec 2025)
- SEC Form 6-K — Hitachi Astemo / Honda
- GPS World — John Deere Gen-2 autonomy kits
- John Deere — Autonomous Tractor page
- Wirtgen Group — John Deere Operations Center
- SEC — ITW Form 10-K
- Honeywell Forge IoT platform
- Yahoo Finance — Honeywell Q3 FY2025 earnings transcript
- Robotics & Automation News — Honeywell sells Intelligrated/Transnorm to American Industrial Partners
- Emerson — DeltaV Mimic Digital Twin Solutions
- Engineer Live — Emerson AspenTech Subsurface Intelligence