Using RViz

RViz (Robot Visualization) is a powerful 3D visualization tool for robotics applications that provides real-time visualization of robot data, sensor information, and system states. The AI Accelerator integrates with RViz to provide comprehensive visualization of AI operations, robot poses, camera feeds, and detection results.

Make sure your robot model and serial number are correctly configured in config/config.yaml then run the following script to start rviz pre-configured to visualize data from the AI Accelerator SDK examples:

./scripts/run_rviz.sh

Some of the most useful ways to use RVIZ:

• Viewing the pose of the camera frames after calibration to confirm that they ended up in the right place (this is also visible under the frames menu in Polyscope X)

• Viewing the collision model for examples that leverage cuMotion. The visualization of the model will automatically update as any changes to the collision yaml file are changed (but changes will not be respected by cuMotion until it is restarted)

• Viewing the poiintcloud from the sensor is valuable in many scenarios - for example using it to see where objects are in the world and aligning collision models with the pointcloud accordingly.

• When using Foundation Pose, the mesh of the object provided to Foundation Pose will also be used as a marker to display the estimated pose in the robot workspace. You can see how this is overlayed on the point cloud and identify any potential issues - for example when the camera is too close to the object for it to estimate depth (<~20cm with the Orbbec 335Lg)