Drift Happens - Advancing calibration-free kinematic estimation with a minimal inertial sensor set

Tracking human movement is valuable for athletic performance and clinical rehabilitation. While traditional optical systems depend on specialized laboratories and markers, wearable inertial measurement units track acceleration and angular velocity in everyday environments. However, their practical adoption remains hindered by the need for calibration routines and sensors across every body segment. This research focuses on enabling motion analysis with minimal sensors and without calibration. To counter orientation drift, where errors in angular velocity measurements accumulate over time, the thesis demonstrates that higher sampling rates, quaternion algebra, and more accurate mathematical approximations improve precision. These methods are supported by a custom sensor system recording raw data at up to 1000 samples per second, along with an open-source evaluation toolbox.