Per https://github.com/ros-perception/point_cloud_transport, we can now use compressed pointcloud streams in ROS 2. We can enable those in the Costmap layers that ingest pointcloud sensor streams (i.e. obstacle, voxel) and the collision monitor and show that this reduces network traffic without composed nodes to keep up better (and CPU impacts).
We would want sensor drivers and depth pipelines to have the option to output this as well to be able to ingest in Nav2 out of the box. Rviz2 + Gazebo would be ideal as well to publish PC2 from Gazebo into Rviz without as much latency
Per https://github.com/ros-perception/point_cloud_transport, we can now use compressed pointcloud streams in ROS 2. We can enable those in the Costmap layers that ingest pointcloud sensor streams (i.e. obstacle, voxel) and the collision monitor and show that this reduces network traffic without composed nodes to keep up better (and CPU impacts).
We would want sensor drivers and depth pipelines to have the option to output this as well to be able to ingest in Nav2 out of the box. Rviz2 + Gazebo would be ideal as well to publish PC2 from Gazebo into Rviz without as much latency
CC @ahcorde