一种基于多传感器的步态相位实时检测系统

对人体步态相位进行准确实时的判断是智能假肢等康复机器人控制的基础,提出并建立了一种实时状态相位检测系统.该系统包含了4只压力传感器和2只姿态传感器,能够准确地区分出平地行走时足跟着地、足放平、足跟离地、足尖离地、摆动弯曲以及摆动伸展等五种状态相位,实验表明:该系统能够适应不同测试者的不同步幅和步速,并能够应用在假肢的实时控制中.     

Gait event detection based on inter-joint coordination using only angular information

The detection of gait events with wearable sensors is necessary for a robotic system interacting with walking people. Conventional gait phase detection methods are based on machine learning. However, this method cannot detect a gait event every gait cycle because it is difficult to extract characteristic points. Additionally, using only angular information for detection is beneficial because angular information is needed for the control and evaluation of the robots. This paper proposes a novel algorithm for the detection of heel contact and toe-off using the inter-joint coordination of the hip, knee, and ankle joints that has a lower-dimensional structure. The proposed algorithm derives the four planes in the angular space and finds the switching points of the planes. Seven participants walked on force plates that measured the force of the foot against the floor. The error was less than 0.035 s when the gait events were detected after calculating planes using the first gait datum. The change in the patterns of the inter-joint coordination reflected the change in gait phases. Although the data were calculated offline, the results show that the heel contact and toe-off could be detected as soon as the angles were sensed once the planes were derived.     

Application of wearable sensors for human gait analysis using fuzzy computational algorithm

The authors have developed and tested a wearable inertial sensor system for the acquisition of gait features. The sensors were placed on anatomical segments of the lower limb: foot, shank, thigh, and hip, and the motion data were then captured in conjunction with 3D ground reaction forces (GRFs). The method of relational matrix was applied to develop a rule-based system, an intelligent fuzzy computational algorithm. The rule-based system provides a feature matrix model representing the strength of association or interaction amongst the elements of the gait functions (limb-segments accelerations and GRFs) throughout the gait cycle. A comparison between the reference rule-based data and an input test data was evaluated using a fuzzy similarity algorithm. This system was tested and evaluated using two subject groups: 10 healthy subjects were recruited to establish the reference fuzzy rule-base, and 4 relapsing remitting multiple sclerosis subjects were used as an input test data; and the grade of similarity between them was evaluated. This similarity provides a quantitative assessment of mobility state of the impaired subject. This algorithmic tool may be helpful to the clinician in the identification of pathological gait impairments, prescribe treatment, and assess the improvements in response to therapeutic intervention.     

一种基于隐马尔科夫模型的步态识别算法

步态是指人体走路时的姿态,步态识别是近年来生物特征识别领域一个备受关注的研究方向。步态阶段的区分是步态识别的重要内容。以隐马尔科夫模型(HMM)为基础,基于安装在膝关节的编码器和大腿部的加速度传感器,在外骨骼辅助行走中识别步态的不同阶段。首先进行数据预处理和特征提取;其次对隐马尔科夫步态识别算法进行设计,包括结构的建立、参数的训练和最终的识别;最后对性能进行评估,总体正确率达到91.06%,说明HMM用于步态阶段识别具有较好的性能。     

Two walking gaits for a planar bipedal robot equipped with a four-bar mechanism for the knee joint

The design of a knee joint is a key issue in robotics and biomechanics to improve the compatibility between prosthesis and human movements, and to improve the bipedal robot performances. We propose a novel design for the knee joint of a planar bipedal robot, based on a four-bar linkage. The dynamic model of the planar bipedal robot is calculated. Two kinds of cyclic walking gaits are considered. The first gait is composed of successive single support phases with stance flat-foot on the ground separated by impacts. The second gait is a succession of finite time double support phases, single support phases, and impacts. During the double support phase, both feet rotate. This phase is ended by an impact of the toe of the forward foot, while the rear foot is taking off. The single support phase is ended by an impact of the swing foot heel, the other foot keeping contact with the ground through its toe. For both gaits, the reference trajectories of the rotational joints are prescribed by cubic spline functions in time. A parametric optimization problem is presented for the determination of the parameters corresponding to the optimal cyclic walking gaits. The main contribution of this paper is the design of a dynamical stable walking gait with double support phases with feet rotation, impacts, and single support phases for this bipedal robot.     

基于隐马尔可夫模型的视频异常场景检测

视频技术的广泛应用带来海量的视频数据,仅依靠人力对监控视频中的异常进行检测是不太可能的。异常行为的自动化检测在公共安全等领域的地位极其重要。提出一种综合考虑目标特性和时空上下文的异常检测方法,该方法利用光流纹理图描述移动物体的刚性特征,建立基于隐马尔可夫模型HMM的时间上下文异常检测模型。在此基础上,提取异常目标的Radon特征,以支持向量机SVM的异常预分类结果为基础,通过HMM建立异常场景的空间上下文分类模型。该模型在公共数据集UCSD PED2上进行了实验验证,结果表明,本算法不仅在异常检测方面优于已有算法,而且还能给出异常分类。     

基于步态序列图像的身份确认

为了利用HMM抽取的步态序列的动态特征来进行身份确认，首先提出一种改进的角度向量用来表征二值化的步态序列图像，以便将每幅图像转化为1维向量，然后再以此作为特征向量，对每个人物建立并训练HMM模型，用于确定人物身份。这种改进的角度向量由于具有较强的抗噪性和方便的尺度伸缩性能，因此既适用于分割质量较差的图像，又能减小行走方向和距离的影响。实验表明，这种HMM不仅能较好地模拟步态的动态特征，还能描述序列图像间的联系，而且算法执行速度快，从输入原始数据到输出识别结果所需时间不超过2min，能满足实时要求。在Soton和NLPR数据库上进行的实验，分别获得了100％和85％的识别率，证明该方法是有效的。     

Fall risk assessment of construction workers based on biomechanical gait stability parameters using wearable insole pressure system

Falls on the same level are a leading cause of non-fatal injuries in the construction industry, and loss of balance events are the primarily contributory risk factors associated with workers’ fall injuries. Previous studies have indicated that changes in biomechanical gait stability parameters provide substantial safety gait metrics for assessing workers’ fall risks. However, scant research has been conducted on changes in biomechanical gait stability parameters based on foot plantar pressure patterns to assess workers’ fall risks. This research examined the changes in spatial foot regions and loss of balance events associated with biomechanical gait stability parameters based on foot plantar pressure patterns measured by wearable insole pressure system. To test the hypotheses of this study, ten asymptomatic participants conducted laboratory simulated loss of balance events which are often initiated by extrinsic fall risk factors. Our results found: (1) statistically significant differences in biomechanical gait stability parameters between spatial foot regions, especially with the peak pressure parameter; and (2) statistically significant differences in biomechanical gait stability parameters between loss of balance events when compared to normal gait (baseline), especially with the pressure-time integral parameter. Overall, the findings of this study not only provide useful safety gait metrics for early detection of specific spatial foot regions but also allow safety managers to understand the mechanism of loss of balance events in order to implement proactive fall-prevention strategies.     

基于行走时脚摆角的步态识别方法

提出一种利用脚摆动特征进行步态识别的方法。对步态序列图像进行背景提取、图像差分、阈值分割、形态学后处理后,提取行走时的脚摆角作为特征参数,再分别采用BP神经网络、最近邻分类器和K近邻分类器法对这些特征数据进行识别分类与比较分析。实验结果表明,与同类方法相比,该方法可以更快速地进行步态识别,且识别性能较好。     

Automated detection of unusual events on stairs

This paper presents a method for automatically detecting unusual human events on stairs from video data. The motivation is to provide a tool for biomedical researchers to rapidly find the events of interest within large quantities of video data. Our system identifies potential sequences containing anomalies, and reduces the amount of data that needs to be searched by a human. We compute two sets of features from a video of a person descending a stairwell. The first set of features are the foot positions and velocities. We track both feet using a mixed state particle filter with an appearance model based on histograms of oriented gradients. We compute expected (most likely) foot positions given the state of the filter at each frame. The second set of features are the parameters of the mean optical flow over a foreground region. Our final classification system inputs these two sets of features into a hidden Markov model (HMM) to analyse the spatio-temporal progression of the stair descent. A single HMM is trained on sequences of normal stair use, and a threshold on sequence likelihoods is used to detect unusual events in new data. We demonstrate our system on a data set with five people descending a set of stairs in a laboratory environment. We show how our system can successfully detect nearly all anomalous events, with a low false positive rate. We discuss limitations and suggest improvements to the system.     

使用运动监视传感网络对行走过程中腿部屈伸角的跟踪

An accelerometry-based gait analysis approach via the platform of sensor network is reported in this paper. The hardware units of the sensor network are wearable accelerometers that are attached at the limbs of human body. For the specific task of gait analysis, flexion angles of the thighs during gait cycles are computed. A Kalman filter is designed to estimate the flexion-extension angle, angular velocity of the thigh using the output of the wearable accelerometers. The proposed approach has been applied to four subjects and the performance is compared with videobased approach. Comparative results indicate that with the proposed Kalman filter, the sensor network is able to track the movement of the thighs during gait cycles with good accuracy and simultaneously detect major gait event of foot contact from the waveform of the angular velocity.     

An unsupervised artificial neural network versus a rule-based approach for fault detection and identification in an automated assembly machine

Artificial neural networks (ANNs) are suitable for fault detection and identification (FDI) applications because of their pattern recognition abilities. In this study, an unsupervised ANN based on Adaptive Resonance Theory (ART) is tested for FDI on an automated O-ring assembly machine testbed, and its performance and practicality are compared to a conventional rule-based method. Three greyscale sensors and two redundant limit switches are used as cost-effective sensors to monitor the machine’s assembly process. Sensor data are collected while the machine is operated under normal condition, as well as 10 fault conditions. Features are selected from the raw sensor data, and data sets are created for training and testing the ANN. The performance of the ANN for detecting and identifying known, unknown and multiple faults is evaluated; the performance is compared to a conventional rule-based method using the same data sets. Results show that the ART ANN is able to achieve excellent fault detection performance with minimal modeling requirements; however, the performance depends on careful tuning of its vigilance parameter. Although the rule-based system requires more effort to set up, it is judged to be more useful when unknown or multiple faults are present. The ART network creates new outputs for unknown and multiple fault conditions, but it does not give any more information as to what the new fault is. By contrast, the rule-based method is able to generate symptoms that clearly identify the unknown and multiple fault conditions. Thus, the rule-based method is judged to be the most feasible method for FDI applications.     

Biomechanical characteristics of slipping during unconstrained walking,turning, gait initiation and termination

Slipping biomechanics was investigated on both non-contaminated and oil-contaminated surfaces during unconstrained straight-line walking (‘walking’), turning, gait initiation and termination. In walking, backward slipping was more frequent, whereas forward slipping was more frequent when turning. Stopping and gait initiation engendered only forward and backward slipping, respectively. Based on slip distance and sliding velocity, severity of forward slipping was least in walking than for the other gait tasks, whereas the tasks had similar effects on backward slipping. Relative to the dry surface, heel and foot contact angles reduced and heel contact (HC) velocity increased for all gait tasks on the contaminated surface. Ground reaction forces were generally lower on the contaminated surface, suggesting kinetic adaptation immediately following HC. Required coefficient of friction (RCoF) did not correlate with slip distance suggesting that RCoF may not be a useful kinetic parameter for assessing slipping risk on contaminated surfaces.

Practitioner Summary: Slipping is hazardous in everyday locomotion and occupational settings. This study investigated foot control kinematics and kinetics across various gait tasks on both a non-contaminated and an oil-contaminated walking surface. Turning, gait termination and gait initiation were associated with a greater risk of slip-related falls than unconstrained walking.     

Adaptive control of an actuated ankle foot orthosis for paretic patients

This paper deals with the control of an active ankle foot orthosis (AAFO) to assist the gait of paretic patients. The AAFO system is driven by both, the residual human torque delivered by the muscles spanning the ankle joint and the AAFO’s actuator’s torque. A model reference adaptive control is proposed to assist dorsiflexion and plantar-flexion movements of the ankle joint during level walking. Unlike most classical model-based controllers, the proposed one does not require any prior estimation of the system’s (AAFO-wearer) parameters. The ankle reference trajectory is updated online based on the main gait cycle events and is adapted with respect to the self-selected speed of the wearer. The adaptive desired ankle trajectory is estimated using cubic spline interpolations between the different key events of the gait cycle. The closed-loop input-to-state stability of the AAFO-wearer system with respect to a bounded human muscular torque is proved by a Lyapunov analysis. Experimental results obtained from three healthy subjects and one paretic patient, show satisfactory results in terms of tracking performance and ankle assistance throughout the full gait cycle. The experiments also show good performance at different walking speeds and with different gait sub-phase duration proportions.     

Two-level control strategy of an eight link biped walking model

This paper presents an adaptive two-level control strategy for a biped walking model and demonstrates its performance in a wide range of walking modes with considerably diverse model and control parameter settings. Proposed control strategy inherits a push off that resembles considerably to forceful extension of the trailing leg during push off in human locomotion and represents a very important source of forward propulsion. Extensive simulations have shown that adjustments in the push off related parameter on higher between-step control level after each step enable evolution of various walking modes of the biped walker at selected walking speeds and distinctive gait patterns. It also allows us to investigate the changes in gait kinematics and kinetics of the biped walking model due to changes in gait velocity, torso inclination and propulsion distribution profiles.     

Effect of silhouette quality on hard problems in Gait recognition.

Gait as a behavioral biometric has been the subject of recent investigations. However, understanding the limits of gait-based recognition and the quantitative study of the factors effecting gait have been confounded by errors in the extracted silhouettes, upon which most recognition algorithms are based. To enable us to study this effect on a large population of subjects, we present a novel model based silhouette reconstruction strategy, based on a population based hidden Markov model (HMM), coupled with an eigen-stance model, to correct for common errors in silhouette detection arising from shadows and background subtraction. The model is trained and benchmarked using manually specified silhouettes for 71 subjects from the recently formulated HumanID Gait Challenge database. Unlike other essentially pixel-level silhouette cleaning methods, this method can remove shadows, especially between feet for the legs-apart stance, and remove parts due to any objects being carried, such as briefcase or a walking cane. After quantitatively establishing the improved quality of the silhouette over simple background subtraction, we show on the 122 subjects HumanID Gait Challenge Dataset and using two gait recognition algorithms that the observed poor performance of gait recognition for hard problems involving matching across factors such as surface, time, and shoe are not due to poor silhouette quality, beyond what is available from statistical background subtraction based methods.     

用块稀疏贝叶斯学习算法重构识别体域网步态模式

针对低功耗体域网步态远程监测终端非稀疏加速度数据重构和步态模式识别性能优化问题,提出了一种基于块稀疏贝叶斯学习的体域网远程步态模式重构识别新方法,该方法基于体域网远程步态监测系统架构和压缩感知框架,在体域网传感节点利用线性稀疏矩阵压缩原始加速度数据,减少传输数据量,降低其功耗,同时在远程终端基于块稀疏贝叶斯学习算法充分利用加速度数据块结构内在相关性,获取加速度数据内在稀疏性,有效提高非稀疏加速度数据重构性能,为准确识别步态模式提供可靠的数据支撑.采用USC-HAD数据库中行走、跑、跳、上楼、下楼五种步态运动的加速度数据验证新方法的有效性,实验结果表明,基于所提算法的加速度数据重构性能明显优于传统压缩感知重构算法性能,使基于支持向量机多步态分类器识别准确率可达98%,显著提高体域网远程步态模式识别性能.所提新方法不仅有效提高非稀疏加速度数据重构和步态模式识别性能,并且也有助于设计低功耗、低成本的体域网加速度数据采集系统,为体域网远程监测步态模式变化提供一个新方法和新思路.     

异常步态3维人体建模和可变视角识别

目的 运用视觉和机器学习方法对步态进行研究已成为当前热点,但多集中在身份识别领域。本文从不同的视角对其进行研究,探讨一种基于点云数据和人体语义特征模型的异常步态3维人体建模和可变视角识别方法。方法 运用非刚性变形和蒙皮方法,构建基于形体和姿态语义特征的参数化3维人体模型;以红外结构光传感器获取的人体异常步态点云数据为观测目标,构建其对应形体和姿态特征的3维人体模型。通过ConvGRU（convolution gated necurrent unit）卷积循环神经网络来提取其投影深度图像的时空特征,并将样本划分为正样本、负样本和自身样本三元组,对异常步态分类器进行训练,以提高分类器对细小差异的鉴别能力。同时对异常步态数据获取难度大和训练视角少的问题,提出了一种基于形体、姿态和视角变换的训练样本扩充方法,以提高模型在面对视角变化时的泛化能力。结果 使用CSU（Central South University）3维异常步态数据库和DHA（depth-included human action video）深度人体行为数据库进行实验,并对比了不同异常步态或行为识别方法的效果。结果表明,本文方法在CSU异常步态库实验中,0°、45°和90°视角下对异常步态的综合检测识别率达到了96.6%,特别是在90°到0°交叉和变换视角实验中,比使用DMHI（difference motion history image）和DMM-CNN（depth motion map-convolutional neural network）等步态动作特征要高出25%以上。在DHA深度人体运动数据库实验中,本文方法识别率接近98%,比DMM等相关算法高出2%~3%。结论 提出的3维异常步态识别方法综合了3维人体先验知识、循环卷积网络的时空特性和虚拟视角样本合成方法的优点,不仅能提高异常步态在面对视角变换时的识别准确性,同时也为3维异常步态检测和识别提供一种新思路。     

Analyzing human gait and posture by combining feature selection and kernel methods

This paper evaluates a set of computational algorithms for the automatic estimation of human postures and gait properties from signals provided by an inertial body sensor. The use of a single sensor device imposes limitations for the automatic estimation of relevant properties, like step length and gait velocity, as well as for the detection of standard postures like sitting or standing. Moreover, the exact location and orientation of the sensor are also a common restriction that is relaxed in this study.Based on accelerations provided by a sensor, known as the ‘9×2’, three approaches are presented extracting kinematic information from the user motion and posture. First, a two-phases procedure implementing feature extraction and support vector machine based classification for daily living activity monitoring is presented. Second, support vector regression is applied on heuristically extracted features for the automatic computation of spatiotemporal properties during gait. Finally, sensor information is interpreted as an observation of a particular trajectory of the human gait dynamical system, from which a reconstruction space is obtained, and then transformed using standard principal components analysis, finally support vector regression is used for prediction.Daily living activities are detected and spatiotemporal parameters of human gait are estimated using methods sharing a common structure based on feature extraction and kernel methods. The approaches presented are susceptible to be used for medical purposes.     

基于足底压力和关节运动信息的步态检测系统设计

作为老年人高发疾病,脑卒中有着较高的几率造成患者行动障碍。及时检测、获取偏瘫患者行走时步态信息,是治疗师为患者制定康复计划的重要环节。目前,较新的步态检测系统大多采用三维分析,三维步态分析系统体积较大、成本高、操作不便,在患者主动康复训练中存在较大的局限。针对现今步态检测系统的局限,设计了一种结合无线传输技术,检测患者关节运动角度及足底压力的运动信息检测系统。系统采用薄膜压力传感器、惯性传感器分别采集足底压力及关节运动信息,实现了对患者的8路足底压力信息及下肢关节运动角度信息的快速、有效采集。实验表明,系统可以有效采集关节运动角度及足底压力信息,在未来的康复训练中有着很好的应用前景。