Skip to content

4. Troubleshooting

Quick reference — jump to the matching symptom below for details:

Symptom Likely cause Section
Reprojection error is high / picks are inaccurate Too few or too similar calibration poses, poor calibration plate visibility Insufficient calibration accuracy
Robot picks above/below the object Height offset or wrong tool (TCP) Height offset in detected poses
No connection / no reply from the camera Wrong IP/port, EthernetKRL misconfigured, robot server inactive Communication errors
Program shows a warning with a negative number (-5001 … -5010) Device-side error, e.g. bad job name or missing calibration Error codes returned by the device
Program exits without a clear error Poses not taught, missing detection pose Program exits unexpectedly

Insufficient calibration accuracy

  • Use more than five calibration poses (seven to eleven give better results). Add CASE blocks in moveTocalibrationPose() and update W_NUM_CALIBRATION_POSES to match.
  • Increase the variation between poses — especially in the pose angles. The variance of the calibration movements matters more than the variance of the poses.
  • Make sure the calibration plate covers as much of the camera image as possible and is fully visible.
  • Prefer a wenglor ZVZJ calibration plate over a printed one (typical reprojection error 0.1 vs 0.5). If printing, print at actual size on stiff, flat material.
  • Check the reprojection error returned by calibration:calculate — high values indicate a poor calibration.

Height offset in detected poses

  • Check that the uniVision job is set properly, especially the height offset from the calibration target to the object in the Device Robot Vision.
  • Ensure the correct tool (TCP) is selected — the program reads the current pose via $POS_ACT.

Communication errors

  • Verify that the network configuration of the wenglor Machine Vision Device matches your setup (default device IP 192.168.100.1, robot server port 6008) in wenglorVision.xml.
  • Make sure wenglorVision.xml was copied to C:\KRC\ROBOTER\Config\User\Common\EthernetKRL\ and that the EthernetKRL (EKI) technology package is installed.
  • Check that the name wenglorVision at W_CONNECTION[] in wenglorUserConfig.src matches the XML file name.
  • Ensure the robot server on the Machine Vision Device is active: device website → Jobs → Processing Instance → Robot Server, with KUKA selected as the robot manufacturer.
  • Check general network connectivity and firewall rules between the controller and the device.

Error codes returned by the device

If the robot server returns a negative error code (-5001 … -5010), the example maps it to a readable message in wenglorGlobal.setReturnError and shows it on the smartPAD via MsgNotify before exiting.

For the meaning of each code, see the Generic Robot Vision Interface → Error codes in the wenglor robot vision manual.

Program exits unexpectedly

  • No calibration poses taught: make sure moveTocalibrationPose() contains a movement for each pose up to W_NUM_CALIBRATION_POSES.
  • Detection pose not taught in moveToDetectObjectsPose() for the camera_not_on_robot use case.
  • No reply from the camera, or an unparseable error code — the program shows a warning and exits. Check the device state via state[<use_case>];.