Local Level Lab

Airborne measurements. Open methods.

Saved calibration-constraint audit

48/48 saved-sector audits pass; state: complete.

Every row preserves all 27 nuisance coordinates. These are generated calibration ranges under necessary constraints, not measured device performance or independent aircraft observations.

Parameter Sectors examined Further narrowed Median ratio to row-contracted width
attitude_x 48 0 1.000000
attitude_y 48 0 1.000000
attitude_z 48 0 1.000000
velocity_x 48 0 1.000000
velocity_y 48 0 1.000000
velocity_z 48 0 1.000000
position_x 48 7 1.000000
position_y 48 6 0.999999
position_z 48 0 1.000000
gyro_bias_x 48 0 1.000000
gyro_bias_y 48 0 1.000000
gyro_bias_z 48 0 1.000000
accel_bias_x 48 0 1.000000
accel_bias_y 48 0 1.000000
accel_bias_z 48 0 1.000000
gyro_gain_x 48 0 1.000000
gyro_gain_y 48 0 1.000000
gyro_gain_z 48 0 1.000000
accel_gain_x 48 0 1.000000
accel_gain_y 48 0 1.000000
accel_gain_z 48 0 1.000000
lever_x 48 0 1.000000
lever_y 48 0 1.000000
lever_z 48 0 1.000000
gravity_delta 48 0 1.000000
time_offset 48 0 1.000000
earth_removal 48 0 1.000000
Case Sectors Existing nonlinear floor exceeds target N / E / height
case-0-0-400 8 0 / 0 / 8
case-0-1-400 8 0 / 0 / 8
case-1-0-400 8 0 / 0 / 8
case-1-1-400 8 0 / 0 / 8
case-2-0-400 8 0 / 0 / 8
case-2-1-400 8 0 / 0 / 8

A floor applies to the retained computed remainder, not to true aircraft uncertainty. Recomputing a rigorously smaller remainder could change it; a better linear optimizer alone cannot erase it. LP failure uses the qualified full-box fallback. No rank threshold drops unidentified columns.

Reserved/used identities: 48/48; completed LP proposals: 2592. Zero native integration, Earth-model fits or original-file reads.

All parameter intervals · Frozen scope · Source comparison · Roadmap.