Drift-Free Limb Angles Reconstruction for FSM-Based Gait Phases Supervision

Synchronizing assistance in soft exosuits requires precise gait phase detection derived from reliable kinematic data. This paper presents a six-phase Finite State Machine utilizing a unilateral modeling approach, which treats each limb as an independent system to ensure a lightweight and modular control architecture. To achieve robust orientation estimation, we implement a signal processing pipeline featuring gravity-based orientation mapping and trajectory unwrapping to manage mathematical singularities. A key contribution of this work is the integration of recursive high-pass filtering for drift suppression, enabling the reconstruction of accurate, driftfree limb angles from Inertial Measurement Unit data. The FSM utilizes deterministic transition logic-relying on threshold analysis and zero-crossing detection-to identify fundamental gait events such as foot impact, heel-off, and swing dynamics. Experimental results demonstrate that the system accurately supervises gait progression with high reliability, providing a scalable and computationally efficient foundation for the control of wearable robotic systems across varied locomotor activities.