ipa-dmo

The PR https://github.com/ros/filters/pull/80 in the filterchain base class prevents the rclcpp::exceptions::ParameterAlreadyDeclaredException based on the filter name and type declarations. For the different parameter of the single grid map filters, a...

Yes, the Pipeline Sequence does not make much sense here. This XML should only show a minimal example of switching planner IDs. The actual BT XML loops over a goal...

I had no time yet to look into it again, will do it later today or tomorrow.

I added some debugging output on my test branch. The `planner_id` API looks fine. For me it looks like some internal issues in the A* planner implementation. The BT selects...

Changing the `tolerance` to `0.0` for both planner configs does not change the behavior @SteveMacenski.

Did some more debugging and looks like the motion_model in the motion_table object of the NodeHybrid Class stays a Reeds-Shepp model after selecting the Dubin planner again: (in the 3rd...

Yes that sounds like the reason. When moving it outside the the ``if`` block it works as intended. In my examples I ran the Hybrid A* Planner with Dubin ``GridBased``...

The first solution with ``thread_local`` does not work as far as I tested, since with every ``ComputePathToPose`` action a new thread is created. So already within one planner id (e.g....

I added some tests to show the behavior of the scale_velocity feature. In general there are two main expections from updating to the commanded curvature: - When we are slowing...