Follow-up from "370 autoware auto fixing nan in result"
The following discussion from !54 (merged) should be addressed:
-
@christopher.ho started a discussion: There should probably also be a check here to ensure that the
heading
field of all trajectory points are valid 2D quaternions (i.e. the norm is approximately 1.0)E: ... in which case you would return
false
since it is not a valid trajectory that can be handled by this controller. As an example, a purely position-based controller (e.g.PurePursuit
) wouldn't care about the heading field.