Small Part Pick and Place Custody Test
A small part pick and place custody test needs more than one gripper bit. Reliable handling requires separate evidence that the part left the pickup nest, stayed with the tool through transfer, left the tool during release, and arrived in the correct destination pose.
EVST uses the Three-Signal Custody Test to keep part ownership explicit while the component moves between stations. It is aimed at automation engineers working with small, reflective, oily, porous, flexible, or easily doubled parts. The method does not assume that vacuum, finger position, or vision confidence means the same thing for every component and environment.
Small part pick and place custody test: name the owner
At every transition, identify whether the part is owned by the infeed, the gripper, the destination fixture, or an exception route. Ownership changes only after evidence. Before pickup, the infeed confirms availability and a valid pose. At lift-off, the cell proves that the selected component left its pocket and is retained by the tool. During transfer, custody remains with the gripper. After release, the destination must confirm presence and any required orientation before the tool is allowed to leave. This simple ledger prevents an uncertain part from disappearing between control steps.
| Decision state | Evidence | Action |
|---|---|---|
| Pickup proof | Part has left the source and is held by the tool | Transfer may begin |
| Retained proof | Custody survives the intended motion and orientation | Continue toward the destination |
| Release proof | Part is no longer attached to the tool | Allow tool withdrawal only with a known destination state |
| Destination proof | Correct pocket contains the expected part and pose | Close the cycle or route uncertainty |
Define the pickup window with real incoming variation
Characterize orientation, overlap, spacing, height, surface condition, contamination, and presentation error using representative parts. A vision coordinate may be accurate while the gripper has no safe contact area, and a part-present signal may be true for a doubled or tilted component. Establish the range where the tool can approach, make the intended contact, close or establish vacuum, and lift without disturbing neighbors. Conditions outside that range should produce a no-pick decision or upstream correction rather than a speculative robot search.
Prove pickup at lift-off
Grip command complete is not pickup proof. Mechanical fingers may close without a part; vacuum may rise against a blocked cup; a sensor may detect the nest instead of the component. Check custody after the part has separated from its support and before rapid transfer. Evidence can combine finger position, pressure behavior, part detection, mass or force response, or a visual check, but it must be challenged with missing, doubled, misoriented, and poorly sealed samples. The result should distinguish secure pickup, definite failure, and uncertainty.
Keep custody during the motion that stresses it
The highest risk of loss may occur after a successful lift, during acceleration, direction change, vertical travel, or passage near a guard or fixture. Test the intended motion profile with the complete tool, hoses, cables, and realistic contamination. For fragile or position-sensitive components, also monitor slip or rotation that does not cause a total drop. A retained-state check should occur where loss would otherwise remain invisible. The transfer path and receiving station must provide a controlled response if custody disappears.
Separate release proof from destination proof
An open gripper or falling vacuum does not prove that the part left the tool, and a destination sensor can be true before the new part arrives. Define release evidence at the tool and presence or pose evidence at the receiving fixture as separate states. Consider adhesion, residual vacuum, magnetism, static charge, wedging, bounce, and a full destination pocket. The next pick should wait until the controller knows that the previous part is no longer on the tool and that the destination either accepted it or routed the uncertainty.
Signals that require a forced trial:
- The source pocket is empty or contains a doubled part.
- A pickup signal turns true without secure lift-off custody.
- The part slips, rotates, or drops during transfer.
- The tool opens but destination presence or pose is not confirmed.
Recover without losing the custody history
When pickup, transfer, release, and destination signals disagree, stop the sequence at a position where the part can be inspected or safely contained. Preserve the last confirmed state, the conflicting evidence, the attempted part location, and the permitted recovery actions. Avoid automatically opening above an unprotected area or returning to the infeed with an unknown component. Retry limits should reflect the failure mechanism; repeated grasp attempts can scatter parts, damage surfaces, or create double handling. Recovery ends only after part ownership is known again.
Validate the complete flow, not a single successful pick
Acceptance trials should cover incoming variation, depleted and full presentation, missing and doubled parts, grip boundary samples, aggressive transfer segments, occupied destinations, sensor loss, communication interruption, and restart. Record the time spent locating, approaching, proving lift-off, transferring, placing, confirming, and recovering separately. EVST can relate those results to robot reach and payload, tool design, sensing, infeed, downstream geometry, safeguards, and maintenance when representative samples and interface rules are supplied.
Keep the acceptance record tied to the physical state
For this application, the handoff record should preserve part geometry and surface, incoming pose range, separation state, pickup datum, allowable contact, destination datum, and mixed or missing-part cases. It should also name the tested configuration, the observed transition, the acceptance evidence, the unresolved dependency, and the disposition of any uncertain output. That record allows another engineer to repeat the trial after a tooling, fixture, software, material, or interface change instead of assuming that an earlier demonstration still represents the current cell. The same record should connect the physical sequence—locate the incoming part, establish the pickup window, secure and confirm the grip, retain custody through transfer, enter the placement window, release, confirm placement, and route an uncertain part—to the relevant hazard boundary: pinch points, dropped parts, unexpected motion, tool or hose collision, sharp parts, access during mis-pick recovery, and restart with unknown custody. This makes later changes visible instead of silently inheriting an obsolete pass.
Frequently asked questions
Why is one grip signal insufficient?
A single signal usually observes only one physical effect. Vacuum, finger position, or part detection can be misleading at pickup, during motion, or after release. Separate transition decisions make it possible to distinguish secure custody, definite absence, and an unknown part state.
When should pickup be confirmed?
Confirm after the component has left its support but before the motion segment that places high demand on the grip. This timing exposes an empty close, blocked vacuum, double pick, or weak contact while the tool is still in a controlled area.
What proves placement?
Placement requires evidence that the part left the tool and that the intended destination contains the expected component in an acceptable pose. Tool-open feedback alone is not enough, and a destination sensor needs a method that distinguishes the new part from a stale or occupied state.
How should an unknown part state recover?
Preserve the last confirmed custody owner, move only along an approved containment path, and inspect or route the component before another pick. Do not release over an unprotected area or restart from a software step that assumes the part location is known.
Conclusion
The Three-Signal Custody Test turns this application into observable decisions rather than a motion-only demonstration. EVST uses the resulting evidence to connect tooling, interfaces, safeguards, quality disposition, recovery, and cycle segmentation. A project assessment can name the remaining trials, but final performance still belongs to the real part, equipment, environment, process, and acceptance method.
Related reading
- Industrial robot architecture and application range
- Payload, reach, and robot-type selection
- Collaborative robot application planning
References
- ISO 10218-2:2025 — Industrial robot applications and robot cells — integration, commissioning, operation, maintenance, and decommissioning of robot applications and cells.
- OSHA Technical Manual: Industrial Robot Systems and System Safety — robot application components, hazards, risk assessment, risk reduction, and safety-system evaluation.
- NIST Robotic Systems for Smart Manufacturing Program — performance requirements, metrics, test methods, and verification thinking for manufacturing robotics.