Inertial Measurement Unit to Segment Calibration Based on Physically Constrained Pose Generation

Abstract

The accuracy of a motion capture (MoCap) system based on inertial measurement units (IMUs) depends on the IMU-to-segment (I2S) calibration, in which the IMU alignment relative to a body segment is calculated using a reference pose such as a standing T-pose. This study proposes a novel I2S calibration system for a reliable MoCap system. In this system, the reference pose for calibration is constrained by physical objects, and this pose is generated by an optimization-based method which incorporates the IMU measurements with the physical constraints between the body and the object. To demonstrate the system, we estimated the chair-sitting and half-squat motions based on the calibration using a chair and estimated the center-of-mass movements of a rider when riding a motorcycle based on the calibration using a motorcycle. The experiments confirmed the improvements of the motion estimation accuracy by the proposed system, in both chair-sitting and half-squat motion and center-of-mass tracking of a rider. Furthermore, the proposed system enables the I2S calibration using product-use poses.

Keywords

• inertial measurement unit
• calibration
• motion capture
• digital human
• motorcycle