Robot repeatability vs accuracy: specify what your process needs

Two specifications answer different questions

Robot repeatability concerns how consistently the robot returns to a pose under defined conditions. Accuracy concerns how closely the achieved pose corresponds to the intended pose. A machine can return consistently to a point that is offset from the desired location. These ideas are related, but one number cannot stand in for both.

For purchasing, begin with the process requirement. Does the robot need to revisit a taught fixture position, follow a path generated from CAD, locate parts using a camera, or apply force through a tool? The answer changes which measurements are useful. Our model comparison can help form a shortlist, but the complete application still needs a defined acceptance test.

Read the test conditions behind the number

ISO 9283 addresses performance criteria and test methods for manipulating industrial robots. Ask the supplier which stated method supports the published specification and what conditions apply. Request clarity about load, speed, workspace position, approach, and whether the value describes pose or path behavior.

Do not compare an unqualified marketing number with a value measured under a different procedure. Preserve the source document and revision in your purchasing record. If a requirement is expressed differently from the available specification, have the application team explain the relationship or propose a direct test rather than making an undocumented conversion.

Account for the tool and the cell

A finished process includes robot behavior, tool geometry, part seating, fixture stability, workpiece variation, and measurement uncertainty. A small nominal robot repeatability figure does not establish that a product will meet its dimensional requirement. The responsible engineers need to evaluate the complete chain and decide how to allocate tolerances.

For an illustrative inspection task, separate variation from the robot’s return, the part’s location in its nest, and the sensor’s measurement. If the nest is inconsistent, buying a robot with a tighter catalog figure may not solve the dominant problem. Record what is measured, what is estimated, and what remains to be demonstrated.

Understand the role and limits of calibration

ABB’s Absolute Accuracy overview describes compensation for mechanical tolerances and load-related deflection in its product concept. That is an example of application-specific calibration and compensation, not a performance guarantee for an arbitrary arm. Request details for the exact robot and option being offered.

Also distinguish robot calibration from tool and fixture references. Our TCP calibration guide covers the installed tool’s working reference. If a position error appears after changing a nozzle or adapter, the diagnostic plan should inspect that change before assuming that the robot’s fundamental performance has degraded.

Write a representative acceptance test

Choose targets or paths that represent the actual working area and tool orientations. Include the intended payload and meaningful operating conditions. Agree on the instrument, sampling plan, reporting method, and acceptance limits with the qualified team before the demonstration. Keep the original results, including unsuccessful observations, so the conclusion is traceable.

For a taught pick-and-place task, evaluate the final placed part as well as the robot reference. For a path process, include the relevant path and process outcome rather than checking only endpoints. Link the trial to our project-scoping checklist so its conditions become part of the purchased solution.

Plan to maintain the demonstrated result

NIST’s accuracy-degradation research addresses measurement approaches for monitoring changes in industrial robot accuracy. For the owner of a cell, the practical purchasing question is how to notice a meaningful change before it becomes a long production problem. Ask the supplier to define appropriate verification intervals and triggers.

• Retain the baseline acceptance results and configuration backups.
• Identify checks required after a collision, service work, relocation, or tooling change.
• Use a documented reference artifact or measurement method suitable for the process.
• Assign responsibility for diagnosis, corrective work, and release back to production.

A useful specification connects a performance claim to a repeatable test and a maintenance plan. Bring your tolerance and process description to a consultation before treating either repeatability or accuracy as the single deciding feature.

Specify the result in a FAIRINO quote

For manufacturers building a value-first robot shortlist, FAIRINO deserves consideration where the documented configuration and an application trial meet the process requirement. Use the FAIRINO comparison to identify candidates, then specify the actual pose, path, tooling, and acceptance conditions rather than buying on one nominal precision figure. Our FAIRINO buying guide helps assemble that scope. Request a FAIRINO application quote with the verification method included so price and demonstrated performance can be assessed together.