Follow-up on MR !327: how to associate correctly under rotations?
Follow-up on this discussion from !327 (merged).
Currently, we're applying association cuts in local coordinates. @lhuth suggested to widen the search window when the DUT is rotated by "applying the cuts for an un-rotated DUT and projecting this onto the rotated plane". This results in something like
search_window = spatial_cut / cos(angle)
which becomes very large for large angles. (In addition, the special case of a 90deg rotation would need to be handled to avoid dividing by zero).
For now we decided to keep the logic as is - and if a sensor is rotated, the association cut can still be widened "by hand" in the configuration file.