search

MCZhi / DIPP

[TNNLS] Differentiable Integrated Prediction and Planning Framework for Urban Autonomous Driving
https://mczhi.github.io/DIPP/
209 stars 41 forks source link

Error occurs in motion planner optimizer #19

Closed EcustBoy closed 1 year ago

EcustBoy commented 1 year ago

Hi~ author, I try to change the cost terms based on your default design, but when I start run it, during gradient backward step in theseus, the following type of error will occur: There was an error while running the linear optimizer. Original error message: torch.linalg.cholesky: For batch 0: U(X,X) is zero, singular U.. Backward pass will not work. To obtain the best solution seen before the error, run with torch.no_grad() (X means number unfixed number, e.g. 2)

And I found this error always occurs as long as I remove acc_cost or steering_cost , but it runs normally when I remove any other cost terms,

I would like to ask if you have encountered this type of error during research? Is the cost function specifically designed to avoid such error? I'll thanks much for your reply~

MCZhi commented 1 year ago

Hi, @EcustBoy. I haven't run into the problem before but you may try another optimizer such as cholmod_sparse_solver.

EcustBoy commented 1 year ago

Hi, @EcustBoy. I haven't run into the problem before but you may try trying another optimizer such as cholmod_sparse_solver.

Thanks for your advice~ p.s. I also wanna know which version of theseus you use? not sure whether it's related to the library version :-)

MCZhi commented 1 year ago

I am using 0.1.2, should be working fine.