Troubleshooting

LED colors

During the boot process, the LED will change color several times to indicate stages in the boot process:

Table 15 LED color codes
LED color Boot stage
white power supply OK
yellow normal boot process in progress
purple
blue
green boot complete, rc_visard ready

The LED will signal some warning or error states to support the user during troubleshooting.

Table 16 LED color trouble codes
LED color Warning or error state
off no power to the sensor
brief red flash every 5 seconds no network connectivity
red while sensor appears to function normally high-temperature warning (case has exceeded 60 °C)
red while case is below 60 °C Some process has terminated and failed to restart.

Hardware issues

LED does not illuminate

The rc_visard does not start up.

  • Ensure that cables are connected and secured properly.
  • Ensure that adequate DC voltage (18 V to 30 V) with correct polarity is applied to the power connector at the pins labeled as Power and Ground in Table 5. Connecting the sensor to voltage outside of the specified range, to alternating current, with reversed polarity, or to a supply with voltage spikes will lead to permanent hardware damage.

LED turns red while the sensor appears to function normally

This may indicate a high housing temperature. The sensor might be mounted in a position that obstructs free airflow around the cooling fins.

  • Clean cooling fins and housing.
  • Ensure a minimum of 10 cm free space in all directions around cooling fins to provide adequate convective cooling.
  • Ensure that ambient temperature is within specified range.

Reliability issues and/or mechanical damage

This may be an indication of ambient conditions (vibration, shock, resonance, and temperature) being outside of specified range (see Table 3).

  • Operating the rc_visard outside of specified ambient conditions might lead to damage and will void the warranty.

Electrical shock when touching the sensor

This indicates an electrical fault in sensor, cabling, or power supply or adjacent system.

  • Immediately turn off power to the system, disconnect cables, and have a qualified electrician check the setup.
  • Ensure that the sensor housing is properly grounded; check for large ground loops.

Connectivity issues

LED briefly flashes red every 5 seconds

If the LED briefly flashes red every 5 seconds, then the rc_visard is not able to detect a network link.

  • Check that the network cable is properly connected to the rc_visard and the network.
  • If no problem is visible, then replace the Ethernet cable.

A GigE Vision client or rcdiscover-gui cannot detect the rc_visard

  • Check whether the rc_visard’s LED flashes briefly every 5 seconds (check the cable if it does).
  • Ensure that the rc_visard is connected to the same subnet (the discovery mechanism uses broadcasts that will not work across different subnets).

The Web GUI is inaccessible

  • Ensure that the rc_visard is turned on and connected to the same subnet as the host computer.
  • Check whether the rc_visard’s LED flashes briefly every 5 seconds (check the cable if it does).
  • Check whether rcdiscover-gui detects the sensor. If it reports the rc_visard as unreachable, then the rc_visard’s network configuration is wrong. See Network configuration.
  • If the rc_visard is reported as reachable, try double clicking the entry to open the Web GUI in a browser.
  • If this does not work, try entering the rc_visard’s reported IP address directly in the browser as target address.

Too many Web GUIs are open at the same time

The Web GUI consumes the rc_visard’s processing resources to compress images to be transmitted and for statistical output that is regularly polled by the browser. Leaving several instances of the Web GUI open on the same or different computers can significantly diminish the rc_visard’s performance. The Web GUI is meant for configuration and validation, not to permanently monitor the rc_visard.

Camera-image issues

The camera image is too bright

  • If the rc_visard is in manual exposure mode, decrease the exposure time (see Parameters), or
  • switch to auto-exposure mode (see Parameters).

The camera image is too dark

  • If the rc_visard is in manual exposure mode, increase the exposure time (see Parameters), or
  • switch to auto-exposure mode (see Parameters).

The camera image is too noisy

Large gain factors cause high-amplitude image noise. To decrease the image noise,

  • use an additional light source to increase the scene’s light intensity, or
  • choose a greater maximal auto-exposure time (see Parameters).

The camera image is out of focus

  • Check whether the object is too close to the lens and increase the distance between the object and the lens if it is.
  • Check whether the lenses are dirty and clean them if they are (see Lens cleaning).
  • If none of the above applies, a severe hardware problem might exist. Please contact support (Contact).

The camera image is blurred

Fast motions in combination with long exposure times can cause blur. To reduce motion blur,

  • decrease the motion speed of the rc_visard,
  • decrease the motion speed of objects in the field of view of the rc_visard, or
  • decrease the exposure time of the cameras (see Parameters).

The camera image is fuzzy

  • Check whether the lenses are dirty and clean them if so (see Lens cleaning).
  • If none of the above applies, a severe hardware problem might exist. Please contact Roboception (Contact).

The camera image frame rate is too low

  • Increase the image frame rate as described in Parameters.
  • The maximal frame rate of the cameras is 25 Hz.

Depth/Disparity, error, and confidence image issues

All these guidelines also apply to error and confidence images, because they correspond directly to the disparity image.

The disparity image is too sparse or empty

The disparity images’ frame rate is too low

  • Check and increase the frame rate of the camera images (see Parameters). The frame rate of the disparity image cannot be greater than the frame rate of the camera images.
  • Choose a lesser Disparity Image Quality. High-resolution disparity images are only available at about 3 Hz. Full 25 Hz can only be achieved for low-resolution disparity images (see also Table 1).
  • Decrease the Disparity Range and increase the Minimum Distance as much as possible for the application.
  • Decrease the Median filtering value.

The disparity image does not show close objects

The disparity image does not show distant objects

The disparity image is too noisy

The disparity values or the resulting depth values are too inaccurate

  • Decrease the distance between the rc_visard and the scene. Depth-measurement error grows quadratically with the distance from the cameras.
  • Check whether the scene contains repetitive patterns and remove them if it does. They could cause wrong disparity measurements.
  • Check whether the chosen rc_visard variant is correct for the application (see Table 2).

The disparity image is too smooth

The disparity image does not show small structures

Dynamics issues

State estimates are unavailable

  • Check in the Web GUI that pose estimation has been switched on (see Parameters).
  • Check in the Web GUI that the update rate is about 200 Hz.
  • Check the Logs in the Web GUI for errors.

The state estimates are too noisy

  • Adapt the parameters for visual odometry as described in Parameters.
  • Check whether the camera pose stream has enough accuracy.

Pose estimation has jumps

  • Has the SLAM component been turned on? SLAM can cause jumps when reducing errors due to a loop closure.
  • Adapt the parameters for visual odometry as described in Parameters.

Pose frequency is too low

  • Use the real-time pose stream with a 200 Hz update rate. See Stereo INS.

Delay/Latency of pose is too great

GigE Vision/GenICam issues

No images