Robot payload is more than part weight

Show the components behind the mass — method added 15 October 2026

Before accepting a payload figure, ask what is inside it. A workpiece mass, an empty-tool mass and a supplied assembly mass describe different boundaries. The useful purchasing record names the operating state and the components represented by each figure. Otherwise, a neatly calculated total can still omit an adapter or count a tool component twice.

Universal Robots' UR10e SW10_15 maximum-payload instructions, checked 15 October 2026 with no publication date displayed, tell users to consider both gripper and workpiece in a pick-and-place load. The same instructions distinguish mass, centre of gravity and inertia when describing the configured payload. This supports keeping the complete attached assembly in view; it does not turn a mass subtotal into a robot-selection result or a transferable limit for another manufacturer's arm.

Build a component ledger for one named state. Give each row a distinct identity, a mass in a consistent unit and a source reference. Record whether the source represents an individual part or an assembly that already includes other parts. If a supplier's tool value includes its adapter, adding a separate adapter row would double-count it. Ask the supplier to resolve that boundary rather than subtracting an assumed value.

Unknown mass is not zero. Keep the component visible and leave its value unresolved until an appropriate record is available. An explicit zero also needs a defensible source and review; the worksheet permits it as an entered number, not as confirmation that the component has no mass. Use the same care with estimates: a reference to an estimate preserves provenance but does not make it a measurement.

For a hypothetical loaded state, a buyer might enter a tool, an adapter and an object as three separately evidenced components. Their subtotal answers only what those entered masses add up to. It does not prove that the list covers the attached assembly, that the adapter is absent from the tool's quoted mass, or that the values belong to the same revision. Reconcile those questions before carrying the number into application engineering.

KUKA's public Load page, checked 15 October 2026 with no publication date displayed, describes mass, centre of mass and inertias as inputs to static and dynamic load analysis. Its actual analysis requires portal access, which was not used for this review. The public description is enough to establish the worksheet's boundary: adding masses does not combine centre-of-gravity coordinates or inertia tensors, evaluate motion, or replace the relevant manufacturer's assessment.

The local component worksheet below withholds its subtotal when a state name, component name, valid mass or mass source is missing, or when component labels repeat after normalisation. Duplicate-label detection catches a recording problem, not every physical overlap: two different labels can still describe the same assembly. The downloadable record preserves the entries and open questions for qualified review. Keep the existing empty/loaded comparison alongside it, and reopen affected records when tooling, workpiece or revision changes. Neither worksheet certifies completeness, capacity or safety.

Keep the empty and loaded states traceable — method added 8 October 2026

An assessment for the heaviest workpiece can leave the empty return motion unexamined. Ask which tool, object and operating state each record describes before carrying its conclusion into another part of the cycle. The empty gripper still has mass, geometry and inertia; picking up an object changes the attached assembly. A maximum part weight alone cannot explain those states.

Universal Robots' UR10e SW10_15 manual, checked 8 October 2026 with no publication date displayed, asks users to consider the gripper and workpiece together and describes acceleration adjustment in relation to configured mass, centre of gravity and inertia. This is evidence for that manufacturer's stated model and software context. It is not a common payload limit for other arms, nor a guarantee that a proposed motion meets the buyer's cycle target.

The same supplier's SW10_12 script documentation warns that the older set_payload function is deprecated because parameters can become inconsistent and that using it resets the inertia matrix. This illustrates why a saved mass value cannot prove the active load configuration. Do not copy a legacy command from this guide into a robot. A qualified integrator must use the applicable current instructions and verify the installed controller, software and state-selection behaviour.

Create a distinct evidence row for the empty tool and each material loaded state. Name the tool, adapters, object and revision. Attach the mass value and its provenance, the centre-of-gravity frame and units, inertia evidence where required, and the motion boundary used in the manufacturer's assessment. Keep measured, calculated and estimated inputs distinguishable. A missing input should remain an unresolved engineering question, not become a convenient zero.

KUKA's public Load page, checked 8 October with no publication date displayed, describes static and dynamic assessments using mass, centre of mass and inertia, and shareable analysis reports. Preserve a report with the exact robot and input case it analysed. Then keep the separate commissioning evidence showing which state was selected in the installed application. A report is not evidence that the controller actually used those parameters at the relevant point in the cycle.

In a hypothetical pick-and-place cell, the supplier provides a loaded-tool assessment but no record for the empty return. Keep the loaded report; do not discard useful evidence. Mark the return-state gap, identify its owner and request the applicable input and motion assessment. If the object or adapter changes, trace that change to the affected rows rather than editing a single headline mass and leaving old report identifiers attached.

RobotAtom's local two-state comparison exposes missing identity, input provenance, motion scope, activation evidence and qualified review separately for each state. Completing one state cannot clear the other. Its export preserves unknowns and restrictions for handoff. It performs no load calculation, supplies no controller settings and approves neither suitability nor safety. More complex applications need additional states and qualified application review beyond this bounded starter record.

Start with the configured load, not the catalogue number

The part's mass is only one input to robot selection. The useful requirement is the complete load attached to the flange in each operating state: the tool and adapters, the workpiece, the centre of gravity, the inertia and the motion the application requires.

A robot's headline payload can therefore be true while the proposed application remains unverified. RobotAtom treats payload as a hard physical gate tied to the exact configuration and conditions, not as a specification that a price or feature score can offset.

Mass, centre of gravity and inertia answer different questions

• Total mass includes the end effector and the handled object. Universal Robots' current UR10e manual tells users to consider both the gripper and object when calculating pick-and-place payload.
• Centre of gravity records where that combined mass acts relative to the tool flange. The same mass can produce a different load when it is held farther from the flange or in another orientation.
• Inertia describes how the load resists a change in rotation. KUKA's own load-analysis tool asks for mass, centre of mass and inertias and evaluates both static and dynamic capacity.
• The required path, acceleration and cycle time matter too. Universal Robots says some load positions can reduce acceleration, while FANUC says its workcell simulation evaluates payload requirements together with reach, paths and cycle time.

Record every material load state

A pick-and-place cycle does not have one load. At minimum, record the empty tool, the tool holding the lightest and heaviest expected objects, and any state created by changing tools or object orientation. For variable or moving contents, state the operating assumption rather than hiding it in a single maximum figure.

For each state, keep the mass, centre-of-gravity coordinates, inertia where required, tool orientation, relevant path and speed, source of the value, and whether it was measured, calculated, estimated or still unknown. An unknown hard load input should stop comparison until it is resolved.

Ask providers for a load-case result, not a brochure excerpt

• Name the exact robot variant, controller and software version proposed.
• Provide the tool and workpiece load data for every material state, including the coordinate reference and units.
• Check the manufacturer's current load diagram or approved load-analysis method for the proposed mounting and motion conditions.
• Model the actual workcell, tooling, fixtures, paths, clearances and target cycle—not only a favourable static pose.
• State any derating, reduced acceleration, path restriction, special setup or further engineering review required.
• Confirm the final configuration with commissioning tests and agreed acceptance measures before production use.

Treat manufacturer tools as evidence with a defined boundary

Universal Robots, KUKA, ABB and FANUC are authoritative sources for their own product instructions, limits and software claims. They do not independently verify that a different supplier's configured cell will meet a customer's throughput, reliability or safety requirements.

ISO 9283 remains the published international standard for industrial-robot performance criteria and related test methods, but a catalogue performance value or simulation still does not replace application-specific engineering, risk assessment and acceptance testing. RobotAtom records the evidence and its source; it does not certify payload suitability, safety or compliance.