Object trajectory differ from the csv file

[etiennedub]

Hi, The object trajectory seem to differ from the input in the csv file. I try to increase the parameter --trajectory_num_spline_segments as suggest in #14 and and also try to set --trajectory_lambda (#33). Both parameters didn't fix the problem.

From my experimentations, the problem seems to occur when one of the quaternion value cross 0. It might be related to singularities as you mentioned in #14.

I include an experimentation where the problem occur: trajectory.zip In this case, the problem occur when qx cross 0.

Do you have any idea how to resolve this ? Thanks for your work !

[supitalp]

Hello, Thank you for the report. Indeed this seems to be related to singularities with the splines. At this point I unfortunately don't have too much time to investigate a fix. As a temporary fix, is there a way you could apply a relative transformation to your input trajectory in order to avoid crossing qx = 0?

Sorry for that.