A gripper force rating is not proof of a secure grasp

Does the trial still cover the contact surface? — added 13 October 2026

A buyer reviewing an old grasp trial needs a traceable answer to a small question: were the part and fingertip surfaces represented in that trial the ones proposed now? Matching a part number and force setting leaves this unresolved. Ask for the dated trial baseline, the proposed surface condition and the evidence connecting them before carrying the result into a new decision.

Robotiq's Hand-E e-Series specification page, rechecked 13 October 2026, includes the fingertip/part friction coefficient in its friction-payload model and recommends testing that coefficient. Its 26 June 2023 force-setting note, checked the same day, identifies fingertip and object materials among the parameters affecting the force relationship. These are product-specific statements. They do not validate another tool, provide a coefficient for an unknown surface or establish that two conditions are interchangeable.

Build a comparison rather than rewriting the old test record. In the first column, identify what the trial actually represented: sample and finish references, known residue or lubricant, fingertip material and condition, and any relevant cleaning or inspection record. In the second, describe the proposed condition using the same categories. Use dated references that a reviewer can retrieve. If a property was never recorded, mark it unknown instead of reconstructing it from a successful video.

Consider a hypothetical handoff in which the archived trial identifies a dry part, while the proposed process may leave a lubricant film. The old run remains evidence about its recorded dry baseline. The film is an unresolved difference until the responsible reviewers determine what evidence is needed and whether existing evidence addresses it. Do not copy a coefficient from an unrelated material pair, assume a wipe restored the baseline or deliberately create a slip or drop to fill the gap.

Separate observation from disposition. An inspection reference may document what was seen; a run record may show an outcome for a particular configuration. Neither automatically establishes the proposed envelope. For each difference, link the evidence offered, retain adverse or inconclusive observations, and name the reviewer who determines its scope. A disposition can be a narrower proposed use, a request for clarification, an authorised evaluation plan or rejection; the record must not force a positive conclusion.

The existing grasp-retention tool now includes a separate contact-surface review with six fields: trial baseline, proposed condition, differences and unknowns, coverage evidence, reviewer disposition, and owner/date/next action. Its local export keeps blanks visible. This is useful before commissioning discussions because it preserves the reasoning behind an evidence transfer without turning a filled form into a pass score. It calculates neither friction nor payload and sends no robot commands.

Keep the comparison beside the original trial identifiers and complete motion baseline. If the finger, part finish, residue, cleaning process or other relevant surface condition changes later, the responsible owners can reopen the affected question without erasing the earlier result. RobotAtom supplies the record structure, not a cleaning prescription, test authorisation, machinery-safety judgement or certification. Qualified configuration-specific review remains necessary.

Reconcile the run tally before using the result

A handoff can contain a convincing successful clip and still leave the buyer unable to tell what happened across the trial. Which planned runs were executed? Which met the stated retention criteria? Which were invalid, showed another outcome or have no accounted result? Keep those counts beside the run identifiers before using the evidence to decide the next step.

The manufacturer sources explain why settings alone cannot answer that question. Robotiq's dated force-setting note describes a current limit and a configuration-dependent relationship to actual force. Its documented Hand-E specifications distinguish grasp types and require acceleration to be considered. Neither source supplies a universal number of runs that proves retention for every application. The tally proposed here is RobotAtom's record-reconciliation method, not a manufacturer test protocol or a statistical validation rule.

Consider a hypothetical authorised trial planned for twenty runs of one defined surface condition. Eighteen were executed: fourteen retained the part within the stated criteria, two were invalid and one was recorded with another outcome. That leaves one executed run unaccounted for and two planned runs not executed. The fourteen retained runs do not erase either gap. Ask for the missing run reference and the reason for each exclusion, rather than presenting the clip as a result for the complete plan.

Use mutually exclusive classifications for this reconciliation. If one run slipped and later dropped, preserve both observations in the raw record, but do not count it twice across the summary buckets. If the classified total exceeds the executed total, resolve the overlap before relying on the tally. A blank count means unknown; it must not silently become zero. Extra runs beyond the plan also need their amended scope and evidence references retained.

Our existing trial register now retains one condition's planned, executed, retained, invalid and other classified counts with its local export. It exposes unaccounted and unexecuted runs without calculating an acceptance percentage. The responsible reviewer still determines whether the sample population, authorised boundary, criteria and evidence are sufficient. A balanced tally can describe a failed trial, and filling its fields gives no permission to perform hazardous motion or deliberate drops.

Trial-record method added 30 September 2026

Before a buyer accepts a gripping demonstration, agree what the record must establish. A successful pick from one clean sample answers a narrow question. It does not cover another surface condition, a changed finger, a faster trajectory or a recovery sequence. Write down the decision first: which configured operation is under review, and which variations remain outside the evidence?

Two manufacturer sources make the distinction concrete. Robotiq's support note dated 26 June 2023 explains that the documented force setting limits motor current; it is not a direct contact-force measurement, and the relationship depends on the configuration. Its Hand-E e-Series specifications distinguish friction and form-fit payloads, identify the fingertip/part friction coefficient and integrator-selected safety factor, and call for acceleration to be considered. These are product-specific statements, not a transferable safe-load rule.

Use the register linked below to freeze a configuration reference and predefine trial criteria with the responsible integrator. Record the robot, gripper, fingers, part revision, settings and motion program together. Give each planned condition a run identifier so a video, measurement file or inspection record can be matched to the setup that produced it. A settings screenshot alone is not evidence that the part stayed seated or undamaged.

• Nominal condition: identify the sample and complete motion, including placement rather than only pickup.
• Variation: record the agreed surface and dimensional states, their boundaries and which were actually represented.
• Motion and stopping: identify the approved trajectory and stop condition covered by the trial evidence.
• Utility loss and recovery: obtain the manufacturer's instructions and responsible safety review before deciding whether and how a condition may be tested.

Preserve failed and invalid runs in the handoff

Record retained, slipped, dropped, damaged, mis-seated and invalid outcomes separately. If a trial contains mixed results, retain the run-level evidence rather than summarising the group as a success. Link the dated raw record, configuration reference, named reviewer and next action. A complete record can show an unacceptable grasp; document completeness is not an acceptance score.

Consider a hypothetical machine-tending trial with dry samples only. The intended process also exposes parts to oil, but that condition has no approved test record yet. Keep the dry results with their exact baseline, record the oily condition as not tested, and assign the missing evaluation to an owner. Do not infer the second result from the first, borrow a coefficient from a different material pair or deliberately create a hazardous drop to fill the gap.

The next decision might be a revised finger design, another authorised trial, a narrower application envelope or rejection of the proposal. Record that disposition and its limitations. Reopen the relevant evidence when the part, surface, tooling, settings or motion changes; the old reference remains a historical result rather than silently becoming proof of the new setup.

RobotAtom's local grasp-retention register supports this handoff. It stores entries in the current tab and exports a plain-text record with open items visible. It does not calculate allowable payload, prescribe a universal test protocol, certify machinery safety or authorise commissioning. Those decisions stay with the qualified project owners.

A force number describes the gripper, not the finished grasp

A gripper can deliver its published force and still fail to retain a particular part. The finished grasp also depends on where the fingers contact, the part and fingertip surfaces, whether the geometry creates friction or form fit, the direction of gravity, the robot's motion and the external loads applied during the task.

The buyer should therefore ask for a grasp envelope: the permitted part population, configured tool, loading cases, damage limits, failure behaviour and evidence that the exact setup completes the required trajectory. A force rating is one input to that record, not proof of secure handling.

Define the part-to-finger interface

ISO 14539:2000 provides vocabulary and a way to present characteristics for grasp-type grippers; ISO records that edition as confirmed in 2021 and currently under review. That public scope helps teams name gripper characteristics consistently, but it does not verify a product or establish that a proposed grasp will hold a buyer's parts.

• Minimum and maximum mass, dimensions, geometry, centre of mass, stiffness and permitted deformation or marking.
• Surface material and relevant dry, oily, wet, dusty, coated, hot, cold or worn states.
• External or internal grip, friction or form fit, contact locations, available contact area and insertion depth.
• Part presentation, tolerance, seams, flash, labels, flexible features and other variation that can move the contact.
• Fingertip material, texture, compliance, wear limit, cleaning method and replacement rule.

Connect force to the actual gripping point

Do not mix commanded force, effective contact force, force per finger and the arithmetic total across jaws. Preserve the manufacturer's definition, measurement conditions, operating pressure or settings, tolerance and the point along the finger at which the value applies.

SMC's current LEHR selection document uses workpiece mass, friction, finger count and a margin in its manufacturer calculation. It also tells users to check effective force at the actual gripping point, external loads and finger moments. Festo's current HPPF datasheet likewise plots force against gripping-point distance and says external fingers, workpiece loads and acceleration contribute to jaw loading. These are maker-specific sizing instructions, not independently verified results or universal design factors.

Build the complete motion and load case

Robotiq's April 2024 Hand-E manual tells users to consider the complete trajectory, robot acceleration and stopping deceleration, gravity direction, eccentricity and forces applied while placing the object. This is first-party guidance for the documented Robotiq configuration; it does not independently prove a proposed cycle and should not be generalized to other tools.

• Robot pose, gravity direction, path, speed, acceleration, planned deceleration and relevant stop cases.
• Part centre-of-mass offset, gripping-point distance, finger overhang and loads on jaws, bearings and adapters.
• Vibration, impact and forces from insertion, extraction, machining, dispensing, inspection or contact with fixtures.
• Normal motion, rejected picks, retries, recovery and foreseeable utility or communication loss.
• A minimum retention requirement and a maximum allowed contact force, stress or deformation for damage-sensitive parts.

Check stroke, fingers and sensing as separate gates

Robotiq documents different recommended payloads for friction and form-fit grasps on the same Hand-E product and provides object-contact and contact-loss states. Those product features illustrate why grasp type and sensing belong in the requirement, but they remain manufacturer claims until the complete configured operation is tested.

• Usable opening and closing range for every material part state, including approach and release clearance.
• Finger geometry, stiffness, fasteners and permissible forces and moments at the actual offsets.
• Part-present, position, force, current, pressure, vacuum, slip or contact-loss signals and their update timing.
• The exact evidence each signal supplies: contact is not automatically correct seating, sufficient retention or an undamaged part.
• Logic for contradictory signals, marginal readings, retries, rejected parts and operator intervention.

Specify power-loss and recovery behaviour

Record what the exact tool does when electrical power, pneumatic pressure, vacuum, robot motion or communications are lost. A tool may open, remain mechanically locked, retain force temporarily or enter another state. Then define the consequence of a dropped, shifted or trapped part and the approved recovery method.

Robotiq says Hand-E has a power-off brake that blocks finger motion. That is a first-party feature claim for that product, not proof that every held part remains retained in every orientation or fault. ISO/TR 20218-1:2018 provides safety-design guidance for the design and integration of end effectors; it does not certify a configured cell or remove the need for application risk assessment.

Validate the representative grasp envelope

RobotAtom can record the grasp requirements, source boundaries, unknowns, manufacturer calculations and test evidence as distinct items. It does not certify gripper suitability, part retention, machinery safety or regulatory compliance. The responsible buyer, integrator, tool specialist and safety owner must decide what applies and whether the evidence is sufficient for the actual application.

• Freeze the robot, gripper, fingers, adapters, software, settings, part set, environment and trajectory.
• Exercise representative dimensional and surface variation, edge cases, starts, stops and agreed utility-loss behaviour.
• Predefine retained, slipped, damaged, dropped, mis-seated, recovered, intervened and invalid-run outcomes.
• Preserve raw results, configuration identifiers, measurement uncertainty, deviations, reviewer and decision.
• Reopen affected evidence after changes to the part, surface, fingers, settings, motion, load or failure controls.