• active pelvis orthosis
• controllo di medio livello
• esoscheletro
• event detection
• exoskeleton
• gait phase estimation
• hip angle peak
• middle-level control
• ortesi attiva di bacino
• picco dell’angolo di anca
• sincronizzazione
• stima della fase di cammino
• synchronization

In lower limb exoskeletons’ control scenario, synchronization between the delivered assistive torque and user’s locomotion is a crucial issue. This is particularly true when considering subjects with impaired locomotion, such as transfemoral amputees (TFAs). Continuous phase estimation-based control strategies rely on a real-time estimate of the user’s gait phase to enhance synchronization. The primary aim of this work is to improve the robustness and the accuracy of an existing phase estimation-based algorithm for the control of an active pelvis orthosis to assist TFAs during ground-level walking (GLW). The algorithm consists of three subsystems dealing with: event detection, adaptive oscillators-based gait phase estimation and phase error learning. This work focuses on the event detector, which detects the maximum hip flexion angle: the criticalities of the original algorithm were investigated offline and addressed through the implementation of several modifications. Offline simulations on data collected with TFAs performing GLW were performed to evaluate the effectiveness of each of the introduced modifications and of combinations of them. Results show an overall tendency to decrease the variability of the distribution of the errors committed by the peak detection algorithm, which would lead to a more reliable estimated gait phase and thus, to a more regular assistive profile. Future work will regard the online validation of the refined algorithm for peak detection with TFAs.