When Heavy-Payload AMRs Become Assembly Workstations: Split, Build and Rejoin

Flexible transport is becoming part of the assembly process

Mobile robots are increasingly being discussed as production equipment, not only as vehicles that replenish a line. At LogiMAT in March 2026, Dürr demonstrated its ProFleet equipment in an interoperability arena with intersecting routes, load handling and mixed traffic. Its April 2 event report connects that demonstration with the company's established automotive assembly applications. The demonstration shows transport coordination; it does not, by itself, prove an assembly process.

The production opportunity is more specific. A workpiece can remain on a mobile carrier while different operations are performed at successive locations. Products needing extra equipment can leave the common route, visit a specialist cell and return. The factory can then change a product route without rebuilding every connection as a fixed conveyor.

In an April 2025 production interview, Mazda described introducing AGVs into powertrain assembly to accommodate different component mounting positions and electric units. This supports the relevance of adaptable positioning in mixed production. It does not establish that every station can be rearranged freely or that transport software can independently decide the assembly sequence.

For buyers, the important question is what happens when an AGV assembly line stops behaving like one continuous line. A shorter process can overtake a longer one. A quality hold can occupy a powered vehicle. A fastening operation can load the fixture differently from transport. A clear route can still lead to a station that cannot perform the required work.

This article examines those consequences for heavy workpieces that remain on their mobile bases during assembly. It assumes work at an authorized standstill unless another mode has been specifically engineered and validated. The calculations and control proposals are illustrative buyer analysis, not performance results from the manufacturers cited.

Decide what stays with the product

The phrase mobile assembly platform should describe an operating architecture, not simply a flat deck with a high payload rating. Identify what travels with the product: its support fixture, adjustable lift, locating datums, tools, work instructions, power supplies and manufacturing record. Then identify what remains at the station.

A useful historical example is Dürr's September 2021 description of the Lotus Hethel project. It specifies 34 AGVs with lifting tables and stop-and-go operation: car bodies remain on the vehicles, which stop for assembly work before progressing. That example establishes a mobile work-carrier architecture. It should not be presented as evidence that operators may work beside any robot while it travels.

Compare three possible equipment boundaries before selecting vehicles. Each creates a different relationship between assembly time and transport capacity.

The existing guide to EV battery pack line feeding considers robots that can leave loaded carriers at stations. The present case changes that assumption. If a vehicle supports a workpiece through a twelve-minute operation, those twelve minutes belong in its assignment history even though its wheels are stationary.

Retaining one fixture can reduce repeated transfers and preserve a convenient workholding relationship. However, some processes need a fixed foundation, different support points or isolation from wheel and suspension compliance. A hybrid layout may keep the product mobile through suitable operations and transfer it to a stationary fixture for a demanding process.

The specification should also separate workpiece support from personnel access. An operator standing on the floor beside a stopped carrier presents a different application from an operator riding on a platform or working beneath a raised load. Do not allow a general promise of ergonomic assembly to conceal those different requirements.

Make the parked vehicle earn permission to become a workstation

An AGV assembly workstation needs a defined transition from transport equipment to process equipment. Arrival near a station is only the beginning. The correct product, fixture and recipe must be present; the workpiece must be located and supported; the necessary services must be connected; and the applicable protective conditions must permit the operation.

Establish the assembly datum under the process load

Navigation locates the mobile base within its operating environment. Assembly depends on the relationship between the workpiece and the tool. That relationship may require mechanical registration, local measurement, fixture identification or other application-specific methods. A vehicle reaching its target coordinates does not prove that a spindle can enter the intended hole.

Use the site's AMR docking tolerance verification guide for the complete tolerance problem. For mobile assembly, extend the acceptance condition to the loaded process state. Measure after locating, support engagement and clamping, then evaluate whether the relationship remains suitable while the operation applies force.

Selecting a heavy payload assembly AGV requires information beyond the mass it carries. AuE Kassel's MonFlex description explicitly includes assembly on the vehicle with applied forces and moments. That is a useful distinction between a transport rating and a process capability. The published description does not supply a universal permissible force for another supplier's chassis.

For a simple illustrative load-path check, a 240 newton-meter reaction torque resisted through a perpendicular lever arm of 0.30 meters corresponds to an 800-newton force. This is only the relationship between one moment and its equivalent force at that arm. It is not a complete clamp design, wheel-load calculation or allowance for dynamic effects.

The procurement consequence is to request the actual force and moment envelope at the fixture interface. Identify where a torque reaction arm, press load or joining force closes its load path. Station-mounted supports may route part of that load directly to the floor; an unsupported mobile deck may transmit it through the vehicle structure and wheels.

Parts added during assembly can change mass, overhang and center of gravity before the next trip. A partially assembled product can also have fewer restraints than the finished product. Refer those changing configurations to the heavy-payload AMR load-path analysis, and define an authorized travel condition for each relevant production stage.

Specify utilities as part of the process interface

A tooling concept may need electrical power, compressed air, hydraulic pressure, extraction or data connections at the stop. Decide whether those services travel with the carrier or connect locally. Include connection time, contamination sensitivity, inspection access and the conditions required before disconnection.

Onboard equipment adds mass and can consume energy while the vehicle is stationary. A battery estimate based only on travel may therefore miss lift movements, tooling power and controls. Conversely, station-fed power creates an interface whose disengagement must be established before departure. Neither arrangement is automatically preferable.

Keep the operating instructions specific to the product revision and station task. An adjustable lift should present the work at the approved height for the operation, not simply return to the last position used. The release record should preserve any configuration that affects the process result or the subsequent travel envelope.

Separate process completion from permission to move

A proposed station sequence can distinguish arrival, registration, work authorization, result capture and departure release. These are design concepts, not standardized protocol state names. The software implementation should assign an owner and evidence requirement to each transition.

A tightening tool reporting a completed cycle does not prove that the correct fasteners were accepted, the tool is clear, supports are retracted or the area is ready for vehicle motion. Equally, a successful route reservation does not authorize assembly. Process controls, fleet commands and safety-related permissions must retain their respective meanings.

ISO 3691-4:2023 addresses safety requirements and verification for driverless industrial trucks and their systems, including AMRs and AGVs. It is not a universal approval for the combined assembly application. The site's heavy-payload AMR safety assessment provides the broader context for evaluating people, loads and operating conditions.

The departure design should address connected tools, temporary supports, loose components and personnel access through appropriate verified measures. Following an interruption, establish the actual physical and process condition before restoring motion authority. Do not make a generic fault-reset button perform that assessment implicitly.

Balance the branches before trying to restore the build order

A mixed model assembly line can give different variants different routes without giving them unlimited freedom. Each product still has required operations and precedence constraints. A controller may select among qualified equivalent cells; it should not bypass an operation merely because its queue is inconvenient.

Audi's July 2022 modular assembly pilot illustrates variant-dependent routing in interior door-panel preassembly. Panels requiring optional equipment visit the relevant station, while other panels can bypass it. This is a useful routing precedent, not a heavy-vehicle throughput benchmark or evidence that assembly precedence can be ignored.

Consider an original example with a target of fifteen units per hour. Standard products use a four-minute branch operation. Premium products use a separate ten-minute operation. Each branch initially has one cell, one product can be processed at a time, and these simplified cycles exclude failures, changeovers and additional handling.

The ten-minute operation has a theoretical capacity of six units per hour. At a sustained 50% premium mix, demand exceeds that capacity by 1.5 units per hour. Additional mobile bases can hold the growing queue for a while, but they cannot increase the station's processing rate. A larger waiting area cannot make that condition sustainable either.

Possible responses include another qualified premium cell, a validated reduction in its cycle, a different allocation of work or a lower overall release rate. Each changes the production design. Buying transport capacity should follow that decision rather than conceal the overloaded resource.

A second cell creates additional usable capacity only if the resources needed to run it are available. Two positions sharing one operator, torque controller, inspection instrument or constrained utility may not process two products independently. Include staffing, tool availability, recipe qualification and common support services when the supplier calculates the proposed increase. An extra physical bay is not sufficient evidence of an extra production resource.

A short sequence shows where the merge consumes time

Now return to a mix that passes the average screen and examine a local burst. Three builds enter their branch queues four minutes apart: premium build P1 at minute zero, standard build S1 at minute four, and premium build P2 at minute eight. This short sequence need not represent the full shift's mix.

P1 occupies the premium cell from minute zero to ten. S1 uses the standard cell from minute four to eight. P2 waits until minute ten, then uses the premium cell until minute twenty. Assume an additional two minutes from branch completion to readiness at the common downstream station, with no route contention in this simplified example.

For this calculation only, cell entry, departure clearance and changeover add no extra occupancy time, and a departing platform immediately frees its branch cell. Real measured turnover time must be added when those assumptions do not hold.

The downstream station must process P1, S1 and P2 in that order, and each downstream operation occupies it for four minutes. There are no earlier jobs in this example. Admission occurs as soon as both the required build and the station are available.

This example isolates an assembly line sequencing problem from a navigation problem. All three routes can execute correctly, yet the required output order creates waiting and a gap. A different release offset might absorb this particular gap at the cost of delaying the first downstream start. It would not remove the sustained premium-cell overload in the 50/50 scenario.

The governing relationship is simple: the next downstream start is the later of that build's readiness time and the previous downstream start plus its processing duration. Real applications add transfer time, changeovers, quality conditions and other constraints. Use the simplified relationship to expose assumptions, not to replace a production simulation.

Reserve places where waiting cannot trap the next build

Assembly line buffer design must consider retrieval order and occupied geometry, not only the number of rectangles on a layout. If S1 stops in the single approach used by P1, the system may be unable to restore the required order even though an apparently empty downstream station exists.

Define which positions permit independent retrieval, which require another load to move first, and whether the complete workpiece envelope fits during approach and departure. Include maintenance access and the space needed for an approved recovery. A painted location is not usable capacity if placing a product there closes an essential route.

Introduce upstream release control before every permissible waiting position is consumed. Reserve capacity for the next required product and for agreed exception handling. The exact policy depends on the layout and routing graph; blindly reserving a complete end-to-end route can also waste capacity or create resource conflicts.

When a station finishes work but cannot release its platform, it remains blocked for the next job. Model that occupied period explicitly. A simulation that instantly removes completed products will overstate cell capacity and understate the number of powered bases tied up in the process.

Let the product record survive a detour, a hold and a restart

A modular assembly system needs a persistent description of what each product has actually completed. The route is an execution decision; the manufacturing requirements belong to the product and its revision. A detour to another qualified cell should not erase a completed inspection or silently authorize a different recipe.

Maintain separate identities for the product, its fixture and the powered vehicle. They can remain associated for a long period without becoming interchangeable. If a damaged base is replaced under an approved procedure, the manufacturing history must follow the product while the new support and transport configuration is verified.

Assembly process traceability should connect the operation instance, recipe revision, station, relevant tool results and quality disposition to that product. Tool completion alone may not establish acceptance. Missing or conflicting results need a defined disposition, and later corrections should preserve an audit trail rather than overwrite the original event.

Give rework its own route back into production

A held product may need to leave the normal route so that other work can continue. That movement does not release its quality hold. Specify a suitable quarantine or rework position, the conditions for moving the incomplete assembly and the authority required to re-enter normal processing.

Re-entry is more than assigning another destination. Determine which operations remain valid, which need repetition and whether the product retains its original downstream position. A repeated torque operation, for example, cannot be assumed acceptable merely because repeating a transport mission is supported.

Keep the waiting and rework footprint inside the capacity model. If a product remains on its powered base throughout a long investigation, it removes that base from normal circulation. A reserve plan might use qualified replacement equipment or a designed transfer to another support. It should not depend on an improvised lift of an uncertain heavy assembly.

Recover physical state before recovering the schedule

After a controller restart, the vehicle may still be located at the station while a tool is engaged or a lift remains raised. Restoring yesterday's “ready” value cannot establish today's safe travel condition. Reconcile current observations with retained process records and apply the approved recovery procedure.

Make the supervisor display explain the blocking condition: an outstanding inspection, unknown fixture support, unavailable downstream position or unresolved process result. “Mission failed” is too coarse when maintenance, quality and production planning have different actions to perform.

The control architecture should identify who authorizes product routing, who confirms station readiness and who releases a quality hold. Those responsibilities may be implemented across MES, station controllers and fleet software in different ways. A successful interface connection does not resolve conflicting ownership of the same decision.

Ask the supplier to demonstrate a production window, not a vehicle lap

AMR assembly line integration should be accepted against completed manufacturing work within the agreed product mix. A robot circulating around an empty demonstration loop cannot establish process capability, output sequence or recovery performance under station blocking.

Specify a representative production window with defined starting conditions, workpiece variants, required operations and ending inventory. The supplier should state what has been measured, what has been simulated and what remains an assumption. Mixing those evidence types into one throughput number makes the quotation difficult to assess.

Use the AMR simulation validation guide to examine model credibility. For this application, require finite station capacity, actual blocking behavior, variant-specific work, merge retrieval constraints and the platform's continued occupation during assembly. Test a range of product mixes and bursts rather than one favorable average.

Travel speed should not be the only adjustable parameter. Ask whether the predicted result changes more when the premium operation is shortened, an independently accessible holding position is added, release timing is revised or another mobile base is introduced. That comparison identifies the resource that actually limits output.

For the physical acceptance run, use approved test methods and authorized load configurations. Include nominal production first, then controlled challenges chosen for the intended process:

1. Run a short cluster of long-cycle variants and observe branch queues, platform occupation and downstream gaps.
2. Make an approved downstream waiting position unavailable and verify that upstream release responds before access is obstructed.
3. Introduce a simulated missing process result and confirm that the product cannot become accepted merely because transport completed.
4. Exercise the approved recovery from an interrupted operation, including physical-state checks and treatment of previously recorded results.
5. Route a designated test product through the rework procedure and demonstrate controlled re-entry without losing its history.

Report accepted output and sequence compliance together. Also show time spent processing, waiting for a cell, blocked after processing, held for quality, recovering and travelling. These categories should account for the relevant platform assignment time without hiding work in an unexplained “idle” total.

Do not end the run at the last dispatched vehicle. Record unfinished assemblies, occupied stations and platforms stranded away from their next usable state. A burst that empties the input buffer while leaving a growing queue before final inspection does not prove sustainable output.

Agree the commercial performance boundary with all purchasing stakeholders. Manufacturing engineering owns the process requirements; quality defines accepted results; operations confirms staffing and intervention work; controls engineering resolves command ownership; procurement makes the delivered scope and performance commitments comparable.

The supplier scope should identify who provides fixtures, lifts, tools, local registration, utilities, guarding, station software, production-system interfaces and recovery equipment. Include the work required when a new product variant arrives. A reusable base is only one reusable element; an incompatible fixture or unqualified tool can still make a new model expensive to introduce.

Retain the evidence necessary to change the layout later: station interface definitions, approved operating modes, routing constraints, configuration versions and the tests affected by each change. Moving a cell can alter cable connections, floor support, approach geometry and human access even when its production recipe stays the same.

Price the next model change while the layout is still on paper

Flexibility has commercial value when the factory can make an anticipated change with a known engineering scope and production interruption. A supplier demonstration of a new route does not establish the cost of qualifying a new workpiece, tool or assembly operation.

Request two change scenarios with the initial bid. In the first, a new variant retains the existing support interface but requires an additional specialist operation. In the second, the product changes its mounting geometry or work envelope and needs a revised fixture. Ask suppliers to identify the hardware, programming, process qualification, training and site verification required for each scenario.

The resulting comparison should distinguish reusable equipment from reusable approval evidence. A mobile base may remain suitable while its fixture, route clearance or stopping envelope needs reassessment. A station recipe may be reusable while a different tool orientation introduces a new reaction load. Count the work that makes the revised configuration usable, not just the components retained.

Also show the production consequences of making the change. Can one qualified cell continue operating while another is modified? Where will unfinished products wait? Which vehicle configurations remain interchangeable during the transition? What conditions allow the previous setup to be restored if the new variant fails its acceptance checks?

These answers give procurement a basis for comparing lifecycle support and expansion options without assuming that every future model can be introduced through software alone. They also give manufacturing engineering an actionable boundary: the changes already covered by the accepted design, and the changes that require new evidence before production release.

Sources and evidence notes

• Dürr: LogiMAT – Experience the logistics of tomorrow live, April 2, 2026, reporting the March 24–26 exhibition. Evidence of the transport demonstration, not an assembly acceptance result.
• Mazda: The ambition behind Monozukuri Innovation 2.0, April 4, 2025. Official account of adaptable AGV positioning in powertrain assembly.
• Dürr: Flexible assembly concept for Lotus, September 2, 2021. Historical description of stop-and-go final assembly with mobile work carriers.
• AuE Kassel: Logistics and conveying. Supplier description of MonFlex assembly platforms and process-force capability; project-specific ratings require separate evidence.
• Audi: Modular Assembly, July 26, 2022. Interior door-panel preassembly pilot illustrating variant-dependent routing.
• ISO 3691-4:2023. Public scope information for driverless industrial trucks and their systems; not a substitute for evaluating the combined assembly application.

Sources checked September 28, 2026. Numerical examples and proposed acceptance methods are editorial analysis. They are not supplier guarantees, reported customer outcomes or a complete machinery design.

#MobileAssembly #AGVAssembly #HeavyPayloadAMR #MixedModelProduction #AssemblySequencing #ManufacturingEngineering #FactoryAutomation #ProductionTraceability