Anisotropic speed (direction-dependent speed)?

70 views
Skip to first unread message

Stephan

unread,
Aug 9, 2014, 5:12:03 PM8/9/14
to sciki...@googlegroups.com
Dear Jason (and others),

Thank you for the excellent module, I love to play around with the fast marching method and scikit-fmm gives me an easy way to explore it.

I thought about whether it is possible to use direction-dependent speed for skfmm.travel_time (e.g. having a slanted floor for robotic indoor navigation). By reading the scientific literature, I understand that the standard algorithm does not apply, since the it updates along the gradient direction (see e.g. http://www2.cvl.isy.liu.se/ScOut/Theses/PaperInfo/qf03.html, p. 84ff). However, it is possible to modify the fmm algorithm to handle anisotropic speed, as explained in http://www2.cvl.isy.liu.se/ScOut/Theses/PaperInfo/qf03.html.

My question: Has anybody considered implementing this? My knowledge of C (and the underlying math) is probably not good enough to attack this on my own.

Thanks again
Stephan

 

Jason Furtney

unread,
Aug 12, 2014, 5:58:16 PM8/12/14
to sciki...@googlegroups.com
Dear Stephan,

Thanks for the email with the link to the thesis. I had not seen this
document before, it looks like a good resource. In principle the
anisotropic methods described could be implemented in sckit-fmm, I had
not considered this. I have a back log of other things to do with the
code first, but I will consider it for a future development. If you
want to try implementing it I can help, just fork the code on github
and start hacking away!

Thanks,
Jason
> --
> You received this message because you are subscribed to the Google Groups
> "scikit-fmm" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to scikit-fmm+...@googlegroups.com.
> To post to this group, send email to sciki...@googlegroups.com.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/scikit-fmm/25a2637d-a726-43c9-a915-29a22ded1ebb%40googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.

Stephan

unread,
Aug 22, 2014, 8:28:31 AM8/22/14
to sciki...@googlegroups.com
Dear Jason,

Thank your for kind reply. I will give it a try, but I have to say that my C capabilities are a bit rusty and I have never done anything with C for Python. Additionally, I'm not feeling completely at home with the math. I will probably come back with questions later on.

Thanks
Stephan
Reply all
Reply to author
Forward
0 new messages