Page cover image

Smart Vacuum Picking Skill

Surface interpretation for vacuum picking.

Overview

A vacuum gripper belongs to the group of astringent grippers in which a continuous holding force is applied without compressive stress. By creating a vacuum inside the gripping system, objects are attracted and held onto the gripper accordingly if the vacuum is maintained sufficiently. The Smart Vacuum Picking Skill is designed for single-suction vacuum grippers.

Depending on the setting of the skill, a certain number of possible pick points is displayed in the camera image. The color coding provides information on the interpretation of the pick points. Points marked in yellow represent potential pick points that also meet the quality requirements of the system. The blue point, on the other hand, defines the best pick point selection, which best meets the requirements. The selection and sorting of the pick points can be set via the parameters in the skill. The pick point position and the quality value are also displayed in the Graphical User Interface.

A classic use case for this skill is, for example, singulation in a department store, since this skill can grab a large number of unknown parts. It does not focus on specific parts, but on an optimal grasping point on any object. This means that numerous items, such as pens, packaging, plates and even bags, can be singled out and placed in the shipping carton with just one skill.

Programm application

The skill is already pre-configured in a way that enables it to handle a wide variety of parts. To test the skill, go to "Option" - "Test". There you have the possibility to run the skill and view the predictions of the grip points.

If the skill is to be optimized for the parts, this can also be implemented in the test area. After the skill has been selected, the parameters can be optimized in the lower right corner. After saving the new parameters, the skill can be executed again immediately. The image on the left shows the new prediction of the skill. With the "Parameter Download" button, the parameters can easily be downloaded, so that they can be restored at a later time.

The different recordings of the skill are stored in the "Capture" area.

Skill Result Information

Position

Position of the grasp point relative to the robot’s coordinate system (in m)

Orientation

Rotation of the grasp point relative to the robot’s coordinate system (in m)

Quality

Quality value for the evaluation of a gripper

Specifications

Conditions

Camera mount:

  • Dynamic

  • Static

Camera distance:

35 – 45 cm

Parts dimensions:

2 – 6 cm

Parts material:

matt - low reflection

Specs

Avg. recognition time:

< 0.5 seconds

Supported grippers:

  • Vacuum

  • Magnet

  • Single contact point

Features

  • Single pick-point recognition

  • Generalized to a variety of object types

  • Picking from bulk

Parameter Example

To ensure accurate identification of various types of objects, the skill parameters can be easily adjusted to fit your specific needs. Here are some recommendations to help you find the perfect parameters for your application. For your convenience, we've only included descriptions of parameters that differ from the default settings.

Smart Vacuum Picking Skill for Boxes
  • Used objects: boxes of varying sizes, matt and shiny surfaces

  • Camera distance: 550mm

  • Camera mount: 30° angled

  • Parameters:

    • Min Quality: 0.32

    • other parameters: default values

Technical Parameter Description

Parameter

Name

Maximum deviation to z axis

Parameter

deviation_to_z_axis

Type

float

Description

Maximum angle in degrees of the pick pose's z-axis to the camera's z-axis.

Name

Shuffle range

Parameter

shuffle_range

Type

float

Description

Maximum number of pick points that can be returned. Ignored if < 1.

Name

Margin

Parameter

margin

Type

float

Description

Minimum distance of pick points to the image edges. It is a percentage of image size and the same along all edges.

Name

Z range

Parameter

z_range

Type

float

Description

Maximum allowed distance (in the z direction) between any pick points and the one closest to the camera.

Name

Min quality

Parameter

min_score

Type

float

Description

Minimum quality for pick points, as reported by the AI model. Setting it too high may prevent pick points from being detected.

Detections

Type

pose (Transformation)

Description

Pick pose for the gripper

Type

pixel (list)

Description

Coordinates of the pick point in the image frame

Type

quality (float)

Description

Quality of the original detection, as reported by the AI model.

Type

optimization_angle (float)

Description

Only present if the parameter optimization_orientation is set to true. The parameter represents the rotation angle in radians that was applied to the original pick pose during the orientation-optimization step.

Last updated