Toward Pervasive Gait Analysis With Wearable Sensors: A Systematic Review
Abstract
对当前的定量步态分析技术进行系统的综述，并提出关键指标，来评估现有的通过可穿戴传感器提取步态特征的方法。它旨在突出这一快速发展的研究领域中的关键进展，并概述研究和临床应用的潜在未来方向。
SECTION III. Methods for Extracting Relevant Gait Features From Wearable Sensors
A.文献检索
B. Kinematics
Kinematic information is a well-established set of gait measures in biomechanical analysis.
从惯性传感器获取运动学信息似乎很直观，但是要获得关于人体运动学的准确空间信息仍然很困难：全局失准和integration drift
C.时序特征
D.利用惯性传感器的步速提取
Laudanski et al. [104] reviewed the current research (16 papers in total) on gait speed estimation using inertial sensors, classifying the current gait speed estimation model into three categories: abstraction model (i.e., machine learning approach), human gait model, and numerical integration, shown in Fig. 5.
TODOS. Chen and J. Lach, “Nonlinear feature for gait speed estimation using inertial sensors”, Proc. 8th Int. Conf. Body Area Netw., pp. 185-188, 2013.利用了生物力学提出了新的特征，自动特征选择算法
E. Novel Features Extracted Using Nonlinear Analysis Techniques 步态稳定性和步态复杂性等
F. Kinetics and Muscle Activity
从惯性传感器提取步态特征
从鞋垫压力传感器和EMG传感器提取步态特征
Step Length Estimation Methods Based on Inertial Sensors: A Review
//根据Perry经典的阶段模型，利用传感器的时序信号可以进行步态周期的划分，可以估算出步频，使得步速可以推导出步幅。
由此步速和步幅/步长的估算类似。