A replacement container may arrive shifted or rotated, changing the space available to the gripper. A 3D camera can locate it so the robot controller uses the current wall and rim positions. Learn which geometry matters and when the available measurements are insufficient.
Required geometry
Define the internal space and obstacles relevant to the tool and carried part.
Visibility
Record which boundaries were observed and which were inferred from the model.
Invalid measurement
Request reacquisition or intervention after movement or when geometry is insufficiently supported.
Decide which geometry the robot needs: internal walls, rim, floor or a reference frame for a known container model. External dimensions alone do not establish gripper clearance. Handles, reinforcing ribs and damaged rims can affect the motion.
For known containers, observed geometry can be fitted to a model. Unknown geometry requires a different approach. State the accepted container types and deformation limits; fitting a familiar model must not hide a real discrepancy.
A 3D capture may miss parts of the rim, wall or floor because products obscure them or the sensor measures the surface poorly. A model can infer missing geometry, but that inference remains an assumption. Track the observed support and remaining uncertainty.
Check dimensions, position and orientation for plausibility. Two visible rim segments may provide useful evidence without establishing that a damaged opposite wall still matches the model. Define a rejection path when observations are insufficient.
| Information | What does this mean for the robot? |
|---|---|
| Directly measured wall or rim | The observation supports the position of that visible surface. Account for measurement uncertainty. |
| Hidden geometry inferred from a model | The model supplies an assumed shape. Check that observations sufficiently support that container type. |
| Container moved after imaging | The previous measurement may be stale. Assess whether a new capture is required. |
Transform the detected container pose through a valid camera-to-robot calibration before updating the working region and obstacles. Check whether the capture remains valid after a container change; movement after imaging makes the geometry stale.
Use measurement and calibration uncertainty when assessing clearance. A barely observed edge does not justify a tight path beside a wall. Both the gripper and the carried part need to fit within the validated free space.
Test shifts and rotations within the allowed infeed range. Repeat with an empty container, partial fill and parts close to the rim. Where relevant, include damaged or contaminated containers as a separate category.
Compare estimated geometry with an independent reference. Separate false acceptance from unnecessary rejection. Verify that control logic detects an invalid or stale container measurement and requests reacquisition or operator action.
Mechanical stops can reduce variation and simplify sensing, but require space, suitable containers and confirmation of correct seating. Vision can measure residual variation while introducing acquisition time and exception handling.
Compare total changeover time, stop maintenance, measurement time, permitted container variation and operator interventions. The choice should follow actual logistics and tolerance requirements.
Share container dimensions, permitted position variation, fill conditions and gripper geometry for a focused review.
Discuss your application