TY - JOUR
T1 - Motion Matching Factor-Informed Foot Kinematics/INS Fusion for Pedestrian Autonomous Positioning Under Complex Gaits
AU - Zhang, Ping
AU - Deng, Zhihong
AU - Shen, Kai
AU - Li, Zhe
AU - Dong, Jiesi
AU - Qi, Wenhao
AU - Mao, Rui
N1 - Publisher Copyright:
© 2026 IEEE.
PY - 2026
Y1 - 2026
N2 - Addressing the intermittent failure of traditional zero-velocity update (ZUPT)/inertial navigation system (INS)-based pedestrian positioning under complex gaits, this article proposes an integrated solution - motion matching factor (MMF)-informed foot kinematics/INS fusion method. MMF, embedded within the Kalman filtering framework, is introduced to quantify the alignment between the kinematics model and reality. Leveraging inherent properties of the filter, the innovation nonorthogonality factor (INOF) and residual factor (RF) are constructed as MMF. Subsequently, an online clustering method is proposed to evaluate the matching level based on MMF in real time and adaptively determine the necessity of fusing kinematics information. Instead of focusing on optimizing the fusion detector, this method adopts a simple quasi-zero-velocity detector with a large threshold based on angular velocity modulus. The challenge of identifying false fusion moments is reframed as a fusion decision-making problem grounded in evaluating the matching level between kinematics model and reality. Experimental results demonstrate that under mixed motion states - including slow walking, fast walking, and running - the proposed algorithm maintains high positioning accuracy and computational efficiency while also exhibiting simplicity, ease of implementation, and strong adaptability.
AB - Addressing the intermittent failure of traditional zero-velocity update (ZUPT)/inertial navigation system (INS)-based pedestrian positioning under complex gaits, this article proposes an integrated solution - motion matching factor (MMF)-informed foot kinematics/INS fusion method. MMF, embedded within the Kalman filtering framework, is introduced to quantify the alignment between the kinematics model and reality. Leveraging inherent properties of the filter, the innovation nonorthogonality factor (INOF) and residual factor (RF) are constructed as MMF. Subsequently, an online clustering method is proposed to evaluate the matching level based on MMF in real time and adaptively determine the necessity of fusing kinematics information. Instead of focusing on optimizing the fusion detector, this method adopts a simple quasi-zero-velocity detector with a large threshold based on angular velocity modulus. The challenge of identifying false fusion moments is reframed as a fusion decision-making problem grounded in evaluating the matching level between kinematics model and reality. Experimental results demonstrate that under mixed motion states - including slow walking, fast walking, and running - the proposed algorithm maintains high positioning accuracy and computational efficiency while also exhibiting simplicity, ease of implementation, and strong adaptability.
KW - Foot kinematics
KW - micro inertial measurement unit (MIMU)
KW - motion matching
KW - online clustering
KW - pedestrian autonomous positioning
KW - zero-velocity update (ZUPT)
UR - https://www.scopus.com/pages/publications/105034786584
U2 - 10.1109/TIM.2026.3680181
DO - 10.1109/TIM.2026.3680181
M3 - Article
AN - SCOPUS:105034786584
SN - 0018-9456
VL - 75
JO - IEEE Transactions on Instrumentation and Measurement
JF - IEEE Transactions on Instrumentation and Measurement
M1 - 9517515
ER -