Skip to Main Content
In personal navigation, pedestrian dead reckoning (PDR) systems based on low-cost self-contained sensors exploit the kinematics of human walking, and are well suited for indoor use and in urban canyons where GPS signals are degraded or not available. Considering the electromyography (EMG), which measures electrical potentials generated by muscle contractions from human body, would reflect the muscle activities during human locomotion, this kind of biomedical signal can be utilized to capture human walking characteristics in PDR. The work presented in this paper is the consecutive step of our pilot studies in further developing a novel and robust PDR solution using wearable EMG sensors to measure walking steps. Our PDR solution includes the EMG-based activity classification, step occurrence detection, and step length estimation, as well as the position calculation with the heading from a two-axis digital compass. To avoid step misdetection, two kinds of activities: walking normally and standing still, are classified via the hidden Markov model classifier fed by the sample entropy features extracted from the raw EMG data. Some EMG statistical parameters are also investigated to establish the optimized step length model. To validate the feasibility and effectiveness of this method, several field tests were conducted by a male tester in two experimental sessions, to demonstrate the effectiveness and practicability of the method using EMG to measure walking steps. Furthermore, the results indicate the performance of the PDR solution is comparable to that of the GPS under open-sky environments.