Event

Proposal of the multi-IMU noise-resistant active calibration framework A2I-Calib++

1 reports · 1 sources

Fangyu Jiang et al. proposed A2I-Calib++, a kinematics-driven noise-resistant active calibration framework for multi-IMU systems on legged robot bodies, capable of estimating time offsets, extrinsic rotations and translations, and inertial biases without the need for manual calibration targets or complex parameter tuning. The framework completes calibration through a temporal correlation search followed by analytical spatial and bias estimation, and under a rolling contact model fuses the calibrated shank IMU with joint kinematics to construct a contact-constrained virtual IMU as a calibration reference. Evaluations on simulation and a real quadruped platform show that its noise sensitivity and simulation calibration error are significantly reduced compared with conventional methods.