[uss_qualifier] check more tightly around clustering threhsholds for nominal_behavior #1260
Conversation
…oviderBehavior checks
| max_details_diagonal_m | ||
| * ( | ||
| 1 - 2*DISTANCE_ERROR_TOLERANCE_FRACTION | ||
| ), # details, just below clustering limit |
There was a problem hiding this comment.
@BenjaminPelletier or @mickmis, above, if the diagonal is set to max_details_diagonal_m * (1 - DISTANCE_ERROR_TOLERANCE_FRACTION), mock_uss will return a cluster rather than the details (at least if I can trust the uss_qualifier logs), at least for NetRID v19 (out of time to check for v22a)
Is this expected? (I think not, but before digging I'd rather check)
I'll bring this up at the next weekly meeting (this comment is mostly to provide context during the discussion)
| min_dim = ( | ||
| 2 * self._rid_version.min_obfuscation_distance_m | ||
| - DISTANCE_PRECISION_M | ||
| ) |
There was a problem hiding this comment.
Note: without the epsilon, when running the NetRID v19 scenario, for a requested diagonal of max_details_diagonal_m * (1 - DISTANCE_ERROR_TOLERANCE_FRACTION), the check below will fail.
(might or might not be related to the fact that we get a cluster when we expect details, depending on how clusters are built by mock_uss)
1 participant
Update the checks for clustering of flights when queries exceed the clustering thresholds: the diagonals for requested areas are now closer to the thresholds.