NDTCuda "lm not converged!!"

[espackman-nv]

Hi @koide3,

I was using the NDTCuda method on the first 11 sequences of KITTI (0-10), and in the terminal output, I saw the message "lm not converged!!" along with very poor accuracy (I measured the Average Translational Error of sequence 01 to be 1295.11 meters!!).

Could you point me to where I could learn more about this error message as well as what I can do to improve the performance?

Thanks!

[koide3]

Maybe it suffered from degeneration of LiDAR data. Basically, NDTCuda in this repo is a lazy implementation written in one holiday, and it's not robust compared to the NDT implemenation in PCL. I do not recommend using it anywhere other than (very) feature-rich environments.