Benchmarking & Diagnostics
Benchmarking & Diagnostics
No calibration solve is self-validating from its own reprojection error alone. axon's CalibrationBenchmark cross-validates a configuration, and diagnostics checks whether a captured dataset has enough rotational variety and whether hand-eye methods agree.
CalibrationBenchmark
Cross-validate an intrinsic configuration and, optionally, a full eye-in-hand round trip.
CalibrationBenchmark
Constructor and best practices for cross-validating a calibration configuration.
View →Run
Cross-validate an intrinsic configuration and, optionally, a full eye-in-hand round trip.
View →Cross Validate
Mean reprojection error over held-out folds of a set of images.
View →Hand-Eye Roundtrip
Recover base_T_board from all frames and report the round-trip reprojection error.
View →Diagnostics
Check whether a captured dataset spans enough of SO(3), and whether hand-eye methods agree.
Rotation Variety
Check whether a captured dataset spans enough of SO(3) for hand-eye to be well-determined.
View →Compare Hand-Eye Methods
Re-solve a dataset with every hand-eye method and flag methods that diverge from consensus.
View →Where to Go Next?
Continue to the next tutorial.
Data Collection & Persistence
Drive a robot through a pose sweep and save or load calibration datasets with Axon.
Next tutorial →