I agree that this is an issue. See more context in this thread: #4822 (comment) for the issue and possible solutions.

Anyone in the community open to helping with this? It involves adding the parameter for enforcing path inversions and using the global goal pose vs the path's end pose considering path inversions when MPPI is used with a feasible planner like Smac Planner Hybrid-A*.

@alex-roba , Thanks for creating the issue.

The quickest solution is to revert the faulty MR #4822

@BriceRenaudeau reverting #4822 would reintroduce the bug that it solved :) Instead, you could open a PR with the solution that @SteveMacenski suggested.

Yes, I understand, for me it was a feature, not a bug. 👍

I've also marked it with help-wanted. I've been noticing many of these tickets have had community members jump in to help too, so I've been leaving nice bite-sized things like this open for them to contribute and grow. If this isn't something @BriceRenaudeau @alex-roba @redvinaa can do, at some point I'll address it myself but wanted to give open opportunity to the community to be involved before I did.

Hi! I'm currently working with the Smac Planner Hybrid-A* and encountered this issue while testing the MPPI controller, which led me to switch to the RPP controller instead. I'd like to take a closer look and work on a definitive fix so that the MPPI controller can be used reliably. If you're okay with it, I’ll go ahead and start working on this.

Please do @brayanpa !

Hi! I've opened a pull request with a proposed fix. Although it's a simple solution, I've tested it both in simulation and on the real robot, and it seems to fully address the issues.

@SteveMacenski let me know if this is close to what you had in mind or if any adjustments are needed. I'm happy to make changes and will be monitoring the thread.

Great! I'll take a look early next week - probably Monday