Failure case of far planner in simulation dataset
Opened this issue · 5 comments
A failure case in campus dataset.
As can be seen in the photo(sorry for the low resolution) The goal_point is in green, which should be un-achievable.
But ideally, the behavior of far-planner should navigate to the rest of the un-explored places, trying to find a way out. After it can't find any way out, it could finally stop.
And as we could see in the visibility graph, there're at least three region to be explored, but far planner no longer inject waypoint.
Seems that the reason of such un-expected behavior is caused by local-planner. I don't know the reason why the local planner don't go to the waypoint and stop there.
I encounter such situations several times in the campus dataset.
I also encounter the same problem in tunnel dataset.
Thanks for the question, could you check (visualize) the whole VGraph in Rviz and see if there is any connection to the goal? And we recently fixed a bug for graph construction, so may be good to try out with the latest code.
Why my VGraph like this :
Why my VGraph like this :
I built my map in an environment of about twenty meters also
like this, have you solved it?