Zhidong Meng
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Zhidong Meng has a total of 10 co-authors. Here are the top co-authors by number of shared publications:
- Zhihong Deng
School of Automation, Beijing Institute of Technology, Beijing, China
5 shared publications - Ping Zhang
School of Automation, Beijing Institute of Technology, Beijing, China
3 shared publications - Zhe Li
School of Automation, Beijing Institute of Technology, Beijing, China
3 shared publications - Kai Shen
School of Automation, Beijing Institute of Technology, Beijing, China
2 shared publications - Alex Mihailidis
KITE Research Institute, Toronto Rehabilitation Institute, University Health Network, Toronto, ON M5G 2A2, Canada
1 shared publication - Andrea Iaboni
KITE Research Institute, University Health Network
1 shared publication
Latest Publications
- 22 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 74 (00189456) , 1-12
System-Level Enhanced Pedestrian Autonomous Positioning Method Based on Hierarchical Optimization Under Sparse Reference
To address the issue of positioning accuracy significant degradation in foot kinematics/micro inertial navigation system integrated positioning over long durations, this paper proposes a system-level enhanced pedestrian autonomous positioning (SE-PAP) method based on hierarchical optimization under sparse reference information. The position sequence output of the integrated positioning system is decoupled into the odometer sequence and direction sequence. A SE-PAP model is constructed based on the mechanisms of odometer error and direction error generation. Subsequently, the displacement vector error cost function is established. A hierarchical optimization method is employed to identify and update the SE-PAP model parameters online with minimizing the cost function, thereby enhancing the performance during the autonomous positioning phase. Experimental results show that the proposed method improves positioning performance by up to 65.72% compared with the Traditional Kalman Filter-based pedestrian positioning method and by up to 54.26% compared with the Extended Kalman Filter-based SE-PAP method. The proposed method effectively handles variations in sparse reference information frequency, demonstrating strong adaptability to heterogeneous tester physiologies, environmental conditions, and motion gaits.
- 22 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 74 (00189456) , 1-12
System-Level Enhanced Pedestrian Autonomous Positioning Method Based on Hierarchical Optimization Under Sparse Reference
To address the issue of positioning accuracy significant degradation in foot kinematics/micro inertial navigation system integrated positioning over long durations, this paper proposes a system-level enhanced pedestrian autonomous positioning (SE-PAP) method based on hierarchical optimization under sparse reference information. The position sequence output of the integrated positioning system is decoupled into the odometer sequence and direction sequence. A SE-PAP model is constructed based on the mechanisms of odometer error and direction error generation. Subsequently, the displacement vector error cost function is established. A hierarchical optimization method is employed to identify and update the SE-PAP model parameters online with minimizing the cost function, thereby enhancing the performance during the autonomous positioning phase. Experimental results show that the proposed method improves positioning performance by up to 65.72% compared with the Traditional Kalman Filter-based pedestrian positioning method and by up to 54.26% compared with the Extended Kalman Filter-based SE-PAP method. The proposed method effectively handles variations in sparse reference information frequency, demonstrating strong adaptability to heterogeneous tester physiologies, environmental conditions, and motion gaits.
- 8 May 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Internet of Things Journal
- Vol. 11 (16) , 27509-27517
Coriolis-Based Heading Estimation for Pedestrian Inertial Localization Based on MEMS MIMU
The zero-velocity-update (ZUPT)-aided foot-mounted pedestrian inertial navigation system (PINS) is a powerful, high-precision, and autonomous positioning system for the IOT applications, such as pedestrian indoor and outdoor seamless positioning. The ZUPT-aided PINS always suffers from the heading error, which leads to a high-order divergence rate of the positioning result. This study uses a magnetometer and proposes a Coriolis-based heading estimation (CHE) method to address this challenge. While the magnetometer is capable of directly correcting heading information, it is susceptible to environmental magnetic interference (MI). Based on the Coriolis theory, the CHE method ingeniously leverages angular rate and magnetic measurement to realize the decoupling between the effective magnetic information and MI. Furthermore, the pedestrians need high-precision height estimation, when walking between the multiple floors. This study proposes a height polynomial model based on linear transformation based on the barometer. The proposed method suppresses the long-term drift of air-pressure measurement and improves the height-estimation precision. The above models are integrated into the ZUPT-aided PINS. A linear Kalman filter is designed to fuse the information and suppress the errors of heading and height. At a complex-walking scenes, the experimental result shows that the proposed algorithm achieves higher 3-D average positioning precision. It is 2.897 m (0.145% mileage) under 30-min two-kilometer walking.
- 8 May 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Internet of Things Journal
- Vol. 11 (16) , 27509-27517
Coriolis-Based Heading Estimation for Pedestrian Inertial Localization Based on MEMS MIMU
The zero-velocity-update (ZUPT)-aided foot-mounted pedestrian inertial navigation system (PINS) is a powerful, high-precision, and autonomous positioning system for the IOT applications, such as pedestrian indoor and outdoor seamless positioning. The ZUPT-aided PINS always suffers from the heading error, which leads to a high-order divergence rate of the positioning result. This study uses a magnetometer and proposes a Coriolis-based heading estimation (CHE) method to address this challenge. While the magnetometer is capable of directly correcting heading information, it is susceptible to environmental magnetic interference (MI). Based on the Coriolis theory, the CHE method ingeniously leverages angular rate and magnetic measurement to realize the decoupling between the effective magnetic information and MI. Furthermore, the pedestrians need high-precision height estimation, when walking between the multiple floors. This study proposes a height polynomial model based on linear transformation based on the barometer. The proposed method suppresses the long-term drift of air-pressure measurement and improves the height-estimation precision. The above models are integrated into the ZUPT-aided PINS. A linear Kalman filter is designed to fuse the information and suppress the errors of heading and height. At a complex-walking scenes, the experimental result shows that the proposed algorithm achieves higher 3-D average positioning precision. It is 2.897 m (0.145% mileage) under 30-min two-kilometer walking.
- 18 March 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Internet of Things Journal
- Vol. 11 (12) , 21904-21915
Adaptive Mid-Stance Phase Observer-Aided Pedestrian Inertial Navigation System for Varying Gait Speeds
Accurate and self-contained pedestrian inertial navigation systems (PINSs) are urgently needed for military and civilian wearable Internet of Things applications and location-based services. The zero-velocity update-based PINS faces two challenges of incorrect zero-velocity detection and invalid zero-velocity assumption when the pedestrian randomly changes gait speed and gait cadence, which lead to rapid divergence of position errors and serious underestimation of mileage, respectively. Therefore, an adaptive mid-stance observer (AMSO) is proposed to overcome the challenges above, which consists of a mid-stance phase detector (MD) and a lever-arm velocity observation (LAVO). First, MD establishes a magnetic field that is periodically distorted with the gait cycle to constrain the generalized likelihood ratio detector to reliably detect the mid-stance phase. Second, LAVO infers the pseudo velocity by compensating the lever-arm effect error during mid-stance phase, and its noise is modeled as a first-order Markov process that attenuates with the increase of the stance stability. To eliminate the magnetic heading error caused by distorted magnetic field, a dynamic time warping method is then used to determine the effective intervals of heading observation and correct the heading of inertial navigation. Finally, the colored-noise Kalman filter is used to fuse all observations and correct the positioning results of PINS. The experimental results showed that the proposed method can effectively improve the positioning accuracy despite the gait speed was switched randomly. The strong adaptability and robustness meet the needs of large-scale pedestrian navigation applications.
- 18 March 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Internet of Things Journal
- Vol. 11 (12) , 21904-21915
Adaptive Mid-Stance Phase Observer-Aided Pedestrian Inertial Navigation System for Varying Gait Speeds
Accurate and self-contained pedestrian inertial navigation systems (PINSs) are urgently needed for military and civilian wearable Internet of Things applications and location-based services. The zero-velocity update-based PINS faces two challenges of incorrect zero-velocity detection and invalid zero-velocity assumption when the pedestrian randomly changes gait speed and gait cadence, which lead to rapid divergence of position errors and serious underestimation of mileage, respectively. Therefore, an adaptive mid-stance observer (AMSO) is proposed to overcome the challenges above, which consists of a mid-stance phase detector (MD) and a lever-arm velocity observation (LAVO). First, MD establishes a magnetic field that is periodically distorted with the gait cycle to constrain the generalized likelihood ratio detector to reliably detect the mid-stance phase. Second, LAVO infers the pseudo velocity by compensating the lever-arm effect error during mid-stance phase, and its noise is modeled as a first-order Markov process that attenuates with the increase of the stance stability. To eliminate the magnetic heading error caused by distorted magnetic field, a dynamic time warping method is then used to determine the effective intervals of heading observation and correct the heading of inertial navigation. Finally, the colored-noise Kalman filter is used to fuse all observations and correct the positioning results of PINS. The experimental results showed that the proposed method can effectively improve the positioning accuracy despite the gait speed was switched randomly. The strong adaptability and robustness meet the needs of large-scale pedestrian navigation applications.
- 28 August 2023
- Published by Springer Nature in Biomedical Engineering Letters
- Vol. 14 (1) , 69-78
Undersampling and cumulative class re-decision methods to improve detection of agitation in people with dementia
Agitation is one of the most prevalent symptoms in people with dementia (PwD) that can place themselves and the caregiver’s safety at risk. Developing objective agitation detection approaches is important to support health and safety of PwD living in a residential setting. In a previous study, we collected multimodal wearable sensor data from 17 participants for 600 days and developed machine learning models for detecting agitation in 1-min windows. However, there are significant limitations in the dataset, such as imbalance problem and potential imprecise labels as the occurrence of agitation is much rarer in comparison to the normal behaviours. In this paper, we first implemented different undersampling methods to eliminate the imbalance problem, and came to the conclusion that only 20% of normal behaviour data were adequate to train a competitive agitation detection model. Then, we designed a weighted undersampling method to evaluate the manual labeling mechanism given the ambiguous time interval assumption. After that, the postprocessing method of cumulative class re-decision (CCR) was proposed based on the historical sequential information and continuity characteristic of agitation, improving the decision-making performance for the potential application of agitation detection system. The results showed that a combination of undersampling and CCR improved F1-score and other metrics to varying degrees with less training time and data.
- 28 August 2023
- Published by Springer Nature in Biomedical Engineering Letters
- Vol. 14 (1) , 69-78
Undersampling and cumulative class re-decision methods to improve detection of agitation in people with dementia
Agitation is one of the most prevalent symptoms in people with dementia (PwD) that can place themselves and the caregiver’s safety at risk. Developing objective agitation detection approaches is important to support health and safety of PwD living in a residential setting. In a previous study, we collected multimodal wearable sensor data from 17 participants for 600 days and developed machine learning models for detecting agitation in 1-min windows. However, there are significant limitations in the dataset, such as imbalance problem and potential imprecise labels as the occurrence of agitation is much rarer in comparison to the normal behaviours. In this paper, we first implemented different undersampling methods to eliminate the imbalance problem, and came to the conclusion that only 20% of normal behaviour data were adequate to train a competitive agitation detection model. Then, we designed a weighted undersampling method to evaluate the manual labeling mechanism given the ambiguous time interval assumption. After that, the postprocessing method of cumulative class re-decision (CCR) was proposed based on the historical sequential information and continuity characteristic of agitation, improving the decision-making performance for the potential application of agitation detection system. The results showed that a combination of undersampling and CCR improved F1-score and other metrics to varying degrees with less training time and data.
- 10 December 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 71 (00189456) , 1-9
Magnetoresistive Sensor Error Compensation Method Using Geometry-Constraint Contour Scaling
The magnetoresistive sensor strapped to the projectile is easily disturbed in highly dynamic environments, which has the problem of low accuracy for sensitive geomagnetic signals. In this article, a magnetoresistive sensor error compensation method using geometry-constraint contour scaling is proposed. A benchmark ellipsoid is obtained based on a two-step method. The shortest distance between any measured single point and the benchmark ellipsoid is calculated, and its position distribution relative to the benchmark ellipsoid is determined. The general form of the benchmark ellipsoid is converted into the standard form. The contour is scaled according to the shortest distance, and the single-point ellipsoid is obtained. The calibration parameters are extracted so that the single-point error compensation can compete. Simulation and experiment results show that the compensation accuracy of the proposed method is improved by two orders of magnitude compared with the traditional method.
- 10 December 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 71 (00189456) , 1-9
Magnetoresistive Sensor Error Compensation Method Using Geometry-Constraint Contour Scaling
The magnetoresistive sensor strapped to the projectile is easily disturbed in highly dynamic environments, which has the problem of low accuracy for sensitive geomagnetic signals. In this article, a magnetoresistive sensor error compensation method using geometry-constraint contour scaling is proposed. A benchmark ellipsoid is obtained based on a two-step method. The shortest distance between any measured single point and the benchmark ellipsoid is calculated, and its position distribution relative to the benchmark ellipsoid is determined. The general form of the benchmark ellipsoid is converted into the standard form. The contour is scaled according to the shortest distance, and the single-point ellipsoid is obtained. The calibration parameters are extracted so that the single-point error compensation can compete. Simulation and experiment results show that the compensation accuracy of the proposed method is improved by two orders of magnitude compared with the traditional method.