A Traceable Vectorial Calibration for MEMS Accelerometers Using Industrial Robotic Manipulators

Abstract

This article proposes a systematic, robot-assisted calibration methodology that leverages the traceable kinematics of an ABB IRB-140 industrial manipulator to achieve full spatial observability. Unlike conventional approaches that minimize the scalar norm of the gravity vector, the proposed technique employs a vectorial cost function within the Levenberg–Marquardt (LM) algorithm to independently estimate bias, scale factor, and nonorthogonality parameters. Experimental validation using an MPU-6050 inertial measurement unit (IMU) demonstrates a significant enhancement in sensor fidelity, reducing the standard deviation of angular errors by up to 93% and achieving consistent subdegree precision in the estimated tilt angles. A comparative analysis confirms the superiority of this traceable framework over standard ellipsoid fitting methods, establishing industrial manipulators as high-precision metrological tools for elevating low-cost hardware to the reliability standards required for attitude and heading reference systems (AHRSs).