Hybrid Gait Localization Using Foot-Mounted IMU with SVM Classification

This study proposes a hybrid gait-based localization method utilizing foot-mounted Inertial Measurement Units (IMUs), which implements dynamic threshold adaptation through real-time motion pattern classification and Zero Velocity Update (ZUPT) techniques, while integrating a Kalman Filter (KF)-based positioning framework to achieve human localization under coloured measurement noise (CMN). The method utilizes a support vector machine (SVM) classifier trained on IMU data to accurately distinguish between walking and running gaits, and dynamically adapts ZUPT thresholds and noise parameters according to the identified motion characteristics to accommodate diverse locomotion patterns. The system employs a single foot-mounted IMU combined with KF under CMN (cKF) for trajectory correction. The results of the experiment demonstrate that this approach significantly enhances positioning accuracy in scenarios with mixed gaits.