Abstract—Magnetometer, gyroscope, and accelerometer are
commonly used sensors in a variety of applications. In addition to
sensor’s physical imperfection, magnetometer’s parameters are
also affected by magnetic disturbance. Specifically, the soft-iron
magnetic effect not only changes its intrinsic model parameters,
including the scale factor, orthogonality matrix, and bias, but
also its relative misalignment with respect to other sensors, such
as inertial sensors of interest in this paper. Almost all existing
methods rely on the local gravity information for cross-sensor
calibration, thus requiring to collect accelerometer measurements
at static positions. Based on the rationale that in a homogenous
magnetic field the magnetometer’s measurement variation is
exclusively induced by orientation change, this paper proposes
a novel magnetometer-inertial sensor misalignment estimation
algorithm requiring no local gravity information. Founded on a
constrained optimization, the algorithm is recursive in time with
self-initialization and is also capable of estimating the gyroscope
bias as an added benefit. Field test results show that the proposed
algorithm has quite good estimation accuracy. As it is inherently
immune to any acceleration disturbance, the test equipment does
not have to keep still for effective measurement