How to do Object-Based Calibration

Updated on 2026 / 08 / 27

This article is applicable under the following conditions:

TMflow version: 1.76.6300 or later

TM Robot hardware version: All

Please note that, depending on the TMflow version, the user interface and the operation procedure may differ.

 

 

HOW TO DO OBJECT-BASED CALIBRATION

 

When to do calibration

 

  • Large and small calibration plates are both used to create a workspace during fixed positioning. If any of the calibration plates is > 30 cm away from the camera, causing the camera to produce a low-resolution image of the plate, then the plate cannot be clearly identified. This prevents the user from calibrating the workspace correctly.
  • Object-based calibration is applicable to the eye-in-hand (EIH) setup only, which employs the difference in the robot servoing movement to calculate the relative relationship between the object and the TM Robot, eliminating the need to create a workspace.

  • An object for calibration that has a definite shape and lacks symmetry is easier to position accurately.

     

Causes of failure to create a workspace

 

  • The working distance is > 30 cm.

     

 

 

  • If the camera’s resolutions are set to 1280×920, it produces low-resolution images.

  • If the TM Robot cannot find all points on the calibration plate, click the video link below to check out how this problem can be solved through object-based calibration.

     

Video 1

 

Tilt correction

 

  • Setup:
    • If the working distance is large, the TM Robot cannot find the calibration plate when calibrating a workspace.
    • In this case, manipulate the robot to a distance where it can find the calibration plate.

 

 

  • Create a vision job
  • Select the EIH 2D camera of the robot

 

 

  • Open Camera Kit.
  • Set the camera’s parameters and select an appropriate focus value.

 

 

  • Click Tilt Correction

 

 

  • Enter the UI for Tilt Correction
  • Make sure the camera and the workspace are aligned
  • Execute tilt correction using a supplied calibration plate.
  • Select TM-Board and click Auto-Tilt.

 

 

  • Lifting the robot to the working distance
  • Manipulate the robot along the z-axis direction of the tool base to the working distance.

 

 

Editing the vision flow

 

  • Select Object-based Cali.
  • Enter the editing process for the vision flow.

 

 

 

  • Pattern matching (shape):
    • Click Find > Pattern Matching(shape).

 

 

  • Place distinguishable and target objects at the same height as the same workspace.
  • The distinguishable object should be matched with objects with visible shapes; the features obtained through this matching are directional.
  • The example below displays a magnified TM Landmark as a distinguishable object.

 

 

  • Acquire the edges of TM Landmark and the characters printed on the landmark, and ensure its directionality.
  • Calibration:
    • Finish the Find process and go to Calibration.

 

 

  • Select an appropriate movement range.
  • If the movement range is too small, the system may send a warning “Over the servoing limit.”
  • After the settings are completed, move the TM Robot to its initial position. Then click Start Calibration.
  • The robot moves to five different positions during calibration.
  • Hold the + button on the robot stick until the calibration is finished. This prevents the robot from colliding with any obstacles while on the move.
  • If the features matched are not clear enough, the yellow-colored scope of the actual object sampled becomes inaccurate, as shown in the image below. In this case, make necessary changes in the Find module.

 

 

  • Acquire the object’s base:
    • After the vision flow is edited, execute the flow to obtain the robot base relative to the object—that is, the workspace of the target object.

    • Go back to the Find module, check the target object, and edit the flow for the object.

       

Troubleshooting

 

  • Issue
    • If the TM Robot keeps failing to match an object’s features while it is servoing the object during calibration, a dialog box will appear as follows:

 

 

  • Solutions:
    • Check whether the features matched are too sophisticated. Then go back to the Pattern Matching process and remove any feature that is difficult to match on the pattern matched.
    • Use an Enhance module to make any to-be-matched feature more visible and easier to find.

    • Set a lower score in the Find module. If the pattern is symmetrical, impose a limit on the angle for finding the pattern.

       

Demo with cardboard

 

  • Setup:
    • Object-based calibration is performed using the logo “TMROBOT” printed on the top of a piece of cardboard. Then a flow is edited in ways that manipulate the TM Robot to use a suction nozzle to pick up the cardboard.
  • Tools:

    • Cardboard with a “TMROBOT” logo printed on it

       

       

    • ROBOTIQ Epick Gripper

       

       

  • Workspace:
    • Decide the distance between the camera and cardboard
    • The distance between the camera and cardboard is about 40 cm.
    • The workspace cannot be calibrated through fixed positioning.
      • Execute object-based calibration to obtain the position of the cardboard.

 

 

  • Editing the vision flow:
    • Open Camera Kit.
      • Set the camera’s parameters and focus.
      • Place a calibration plate and do tilt correction.
    • Select Pattern Matching(shape).
      • Circle the TMROBOT logo at the center of the cardboard and the shape of the cardboard.

 

 

  • Do calibration:
    • Set the movement range  
    • start calibration.

 

 

  • Editing the flow:
    • Use a vision job to find the cardboard.
    • Teach the robot where to pick up the cardboard and install a suction nozzle for the robot to fetch the object.

 

 

  • Results:
    • The robot finds the cardboard and arrives at its center.
    • The robot uses the suction nozzle to pick up the cardboard.

  • Demo Video

     

Demo with SSDs

 

  • Setup: Several SSD are placed on their respective trays, object-based calibration performed using the words printed on the top of the SSDs and the shape of the drives, and the TM Robot is manipulated to use a suction nozzle to pick up the drives.

     

  • Tools:

     

     

  • SSD

  • ROBOTIQ Epick Gripper

     

 

  • Workspace:
    • Set up a workspace for the SSDs.
    • The distance between the camera and trays is about 50 cm, so the trays can be entirely seen by the camera.
    • The workspace cannot be calibrated through fixed positioning.
      • Execute object-based calibration to obtain the positions of the SSDs.

 

 

  • Editing the vision flow:
    • Open Camera Kit.
      • Set the camera’s parameters and focus.
      • Place a calibration plate and execute tilt correction.
      • Use an Enhance module to make the SSDs more identifiable.
      • Use Pattern Matching(shape) to circle the words printed on their center and the shape of the drives.

 

 

  • Do calibration:
    • Set the movement range
    • start calibration.

 

 

  • Editing the flow:
    • Use a vision job to find the SSDs.
    • Teach the robot where to pick up the SSDs and install a suction nozzle for the robot to fetch the objects.

 

 

  • Results:
    • The robot finds the SSDs and arrives at their center.
    • The robot uses the suction nozzle to pick up the SSDs one after another from the trays.
  • Demo Video

Was this article helpful ?