13. Limitations and Known Issues

  1. RViz Visualization: RViz visualization is displayed on a remote Ubuntu PC.

  2. Ctrl+C Termination: In some cases, terminating a ROS node or a ROS launch using Ctrl+C can be slow and may escalate to SIGTERM. This behavior is more frequent in the ROS 2 container.

  3. Default Semantic Segmentation Model: The default semantic segmentation model used in ti_vision_cnn and ti_estop nodes was initially trained with the Cityscapes dataset and subsequently re-trained with a small dataset collected from a stereo camera (ZED camera, HD mode). This re-training was done for a limited set of scenarios with coarse annotation. As a result, the model’s accuracy performance may be limited when used with a different camera model or in different environmental scenes.

  4. ROS 2 Demo Launch: Launching a demo in the ROS 2 environment with “ros2 bag play” in a single launch script currently not stable. It is recommended to launch “ros2 bag play” in a separate terminal. The demos in ROS 2 container currently run more stable with live cameras (ZED stereo camera or USB mono camera).