We'd like your input on how you work with planning scenes

[EzraBrooks]

Hi MoveIt community,

As we endeavor at PickNik to improve MoveIt2's support for mobile manipulation workflows, we would like to ask you all how you initially populate the planning scene for your robot. In our experience, the most common ways to populate a PlanningScene at startup are:

• Defining your robot's entire work area in your URDF, especially for fixed-base robots
• Importing a PlanningScene from a .scene file using publish_scene_from_text
• Programmatically constructing your PlanningScene on every bringup in your application code
• Not defining an initial PlanningScene at all, and only populating the scene on-the-fly from perception inputs

The keen-eyed amongst you may have already noticed that we made some improvements to passing URDF planar joints through to OMPL planar joints in the first use case listed above, but defining your world in your URDF doesn't scale well for obvious reasons, so we're moving on to more complex and scalable workflows. As we work with NASA on lunar rovers.. you can't define the entire moon in your URDF 😉

We'd like to know how the community generally initializes their PlanningScene, to help guide enhancements to these features!