Extrinsic Calibration and Spindle Axis Diagnostics of Linear Range Sensors in Rotating Configurations

Obtaining a 3-D 360° reconstruction of environments or reconstructing hidden details of complex objects could be a difficult task or very expensive with current technologies. However, this challenge can be overcome by using a single profilometer in a rotating configuration. In this context, precise calibration of the sensor's pose with respect to the rotating axis is critical for obtaining accurate and reliable data. This work presents a novel algorithm and procedure able to estimate the complete set of extrinsic parameters of a profile sensor rotating around a spindle, using a simple target, and to verify the spindle axis stability while suppressing both sensor and environmental outliers. The method can be exploited for 2-D laser scanners or time-of-flight sensors for centimeter-level accuracy in environmental scanning, as well as for triangulating laser-CCD line profilers for high-precision point cloud estimation with micrometer accuracy. The latter case particularly benefits from th...

Obtaining a 3-D 360° reconstruction of environments or reconstructing hidden details of complex objects could be a difficult task or very expensive with current technologies. However, this challenge can be overcome by using a single profilometer in a rotating configuration. In this context, precise calibration of the sensor's pose with respect to the rotating axis is critical for obtaining accurate and reliable data. This work presents a novel algorithm and procedure able to estimate the complete set of extrinsic parameters of a profile sensor rotating around a spindle, using a simple target, and to verify the spindle axis stability while suppressing both sensor and environmental outliers. The method can be exploited for 2-D laser scanners or time-of-flight sensors for centimeter-level accuracy in environmental scanning, as well as for triangulating laser-CCD line profilers for high-precision point cloud estimation with micrometer accuracy. The latter case particularly benefits from the algorithm's ability to remove outliers originating from the sensor and/or environment, including background interference and moving objects like the mechanisms used to move the target. The algorithm was validated through both simulations and experimental tests. The resulting point clouds achieved an accuracy consistent with the sensor's nominal specification, approximately 0.01 mm. Furthermore, it was demonstrated that it is possible to reveal spindle motion deviations, which can significantly affect the accuracy of the final scan.