The time of replanning.

[ghost]

Hello! I'm really impressed with this work.
Regarding this framework, I have a question to seek your advice. When I run test_overtake.py, I find that the relpanning close loop takes a long time, maybe 90s on my device. Is this normal? My cpu is Intel Core i7-12700.
Thanks in advance for your reply！

[zzx9636]

Thanks for your interest. Yes, it is expected behavior sometimes because the underlying planning algorithm is a naive spline search. I suggest switching it with an optimization-based planner such as ILQR, which fits this framework and can offer your real-time performance.

[ghost]

Thanks for your reply!

[ghost]

Hello! When I try to modify the number of incoming vehicles in test_overtake.py to a random number between 1 and 5, I find that it may have been half an hour without replanning or more.Why is this?
Thanks in advance for your reply！

[zzx9636]

I think it stuck in the planning search, but I am not sure why. Although I have not tried more than one vehicle, its behavior should not be too much different from the single-vehicle case for overtaking, as the ego has no knowledge of how many incoming cars are due to occlusion.