基于多传感器数据融合的人体运动模式识别研究

针对单一传感器在人体运动姿态监测中误差较大的问题,提出了一种基于加速度传感器和陀螺仪数据融合的人体运动模式识别方法;该方法使用陀螺仪输出的人体运动信息对加速度传感器采集到的姿态角信息进行修正,采用卡尔曼滤波算法实现多传感器信息的融合,有效提高了姿态角度测量的准确度;根据人体日常的活动状态构建了基于人体姿态角度特征的隐马尔可夫模型实现人体运动模式的识别;实验表明,该方法比采用单一传感器方法识别的准确率高,可以有效区分不同的日常活动行为。     

基于数据手套与虚拟模型的手臂动作识别方法

为了配合残疾人使用助残假手完成日常活动,提出了一种基于数据手套与虚拟模型的手臂动作识别方法.通过数据手套上的MEMS传感器测量手臂运动数据,结合虚拟交互环境中的手臂模型,完成模型对真实手臂的动作复现,并获取手臂模型的位置数据.而后将传感器加速度、角速度等数据的时域特征与手臂模型位置数据的动作轨迹特征结合起来,使用支持向量机算法对预定义的5种手臂动作进行分类识别,识别率达到99.33％,验证了提出的手臂动作识别方法的可行性.     

基于惯性测量系统的人体运动捕捉技术

通过对人体运动捕捉技术的研究,引入对偶四元数姿态变换方法,利用惯性姿态测量系统捕捉手臂关节的运动角速度、运动加速度、运动轨迹等信息。结合惯性传感器捕捉的人体肢体段的运动学参数,建立虚拟人体棒状骨架模型,实时驱动虚拟人手臂运动。实验结果表明,对偶四元数变换的方法结合了四元数旋转和对偶数平移的特点,避免了在大角度下欧拉角产生的奇异现象;对比图像捕捉系统,惯性运动捕捉系统能够方便获得运动姿态数据,快速建立虚拟人体结构模型。     

基于人体传感和Android技术的运动监测系统设计与实现

为了避免运动过程中和比赛中运动姿态不正确以及运动过量对身体造成的伤害,提出了基于人体传感和Android技术的运动监测系统的设计,在运动过程中对人体运动参数和运动姿态实时监测,通过Android平台实现卡路里消耗计算和运动姿态建模评估,以避免出现运动损伤和体能透支现象。监测系统以MSP430单片机为控制核心,采用MPU6050三轴加速度传感器实时测量人体三轴加速度,利用Android平台计算加速度、速度、里程等运动参数、运动中消耗的卡路里能量,以及人体运动轨迹跟踪、定位;并通过人体运动模型识别算法评估人体运动姿态,实现危险动作预判。实验证明系统能够实现实时运动监测、运动跟踪定位,保护运动者的身体健康。     

利用惯性传感器的人体运动空间轨迹追踪

人体运动的空间轨迹追踪是一种利用传感器技术和计算机技术来分析记录人体的运动过程的方法.为了实现人体运动轨迹的空间追踪,本文设计了一种人体可穿戴式的人体运动捕捉系统,通过佩戴在人体关节点的惯性传感器单元来获取肢体的实时姿态信息.惯性传感器由加速度传感器、角速度传感器和磁力计构成.通过微控制单元获取传感器数据,利用低通滤波和卡尔曼滤波来更新四元数,再将预处理后的数据由蓝牙模块实时发送到电脑端.本文通过对肢体运动的不同角度的实验,证明了利用惯性传感器可以追踪人体肢体、运动的空间轨迹.     

基于加速度传感器的人体前臂动作特征提取

针对人体前臂运动识别困难,识别率低的问题,且为了有效并快速提取人体上肢前臂的运动特征,设计了一种基于加速度传感器的人体前臂动作识别数据采集系统.首先利用加速度传感器来获取的加速度原始数据矩阵,然后通过互补滤波器对加速度原始数据矩阵进行滤波处理并消除重力场作用来获取运动特征矩阵.此系统实现了简单有效的加速度运动特征数据的拾取,为后续的人体前臂运动模式的识别提供了可靠的数据基础.     

基于双重模型的计算机手语动作的实现方法

首先给出了一个新的方法来建立手的模型.这个手的模型由两个模型组成:一个是实现运动的骨骼模型,其决定了运动姿态及运动约束;另一个是表现运动的皮肤(表面)模型,用来表现真实感.其次,根据该模型实现手语的动作,利用该方法实现手语动作的关键技术,包括手及手臂旋转变换的变换矩阵、有向拓扑结构描述骨骼模型、表现运动的皮肤模型的拓扑不变性、以及骨骼模型实现运动和皮肤模型表现运动的方法.该方法具有一定的通用性,为网络环境下实现异语种的手语表达与传输打下了基础.     

基于Poser的个性化人体模型运动仿真技术研究

根据服装工业对数字化动态人体模型的需要,提出了一种建立人体动态模型的新方法.该方法针对特定人体的三维扫描数据,用逆向工程软件Geomagic Studio实现个性化的三维人体NURBS曲面模型的构建,将完成后的人体曲面模型导入Poser软件,利用人体特征点确定关节位置, 建立相应的人体骨骼模型, 并将人体表面模型与骨骼模型相匹配,实现人体运动仿真.实验证明,所建立的人体动态模型在外观形状和运动姿态上具有较高的真实性, 操作简便,易于掌握.     

基于骨骼关键点的散打动作识别与评价方法

为提高散打运动辅助打分的公平公正性,提出一种基于骨骼关键点的散打动作识别与评价方法。首先采集十类散打动作构建实验数据集,获取视频关键帧,对图像进行去噪处理。其次通过基于YOLOv5s-CBAM目标检测的HRNet-DSC-CBAM人体姿态估计方法提取人体骨骼关键点坐标。最后利用ST-GCN动作识别方法进行动作识别,并利用DTW动态时间规整算法完成动作评价。实验结果表明,该方法在自制数据集的10类散打动作中表现出良好的识别效果,可以实现辅助打分功能。     

MEMS加速度传感器在微型特种机器人中的应用

提出一种基于MEMS加速度传感器LIS3LV02DL的特种机器人姿态测控和环境感知方案。阐述了系统硬件组成,分析了传感器姿态测控方法和环境感知原理。为提高加速度传感器工作精度,设计了一种加速度传感器校正方法。试验结果表明：该加速度传感器能较好地感知机器人运动的姿态与方向,并实现机器人着陆姿态调整和外部环境振动信息的监测。     

基于Poser模型的三维人体建模方法

人体真实感建模是虚拟人研究领域的一项重要内容。该文提出一种基于模型的快速人体建模方法。人体几何模型的皮肤层由多边形网格组成的表面表示,骨架层建模采用H-Anim标准,并针对Poser软件输出的人体模型缺乏关节点数据问题,给出一种骨架提取算法。人体运动模型应用机器人运动学方法,由骨架层驱动皮肤层实现。在VC++平台上实现对不同姿势下人体模型的骨架提取,仿真了敬礼过程中手臂的运动,验证了该方法的可行性。     

Constrained rigid body stability and control

Stability and control of a single or three‐body constrained system are considered. Several different types of constrained motion are among them: the impact phase of a free body colliding with the ground, contact with a stationary or moving platform, movement on a frictionless surface or multiple rigid bodies connected by holonomic constraints, and moving as in the human arm. The single body constrained system is controlled by sliding mode control. The stability of the three‐link arm at arbitrary equilibrium points and Lyapunov stability in the vicinity of the equilibrium point are formulated. The formulation and derivations are by computational tools, that is, state space analysis and matrices. The approach can easily be extended to larger systems with many rigid bodies such as skeletal systems. The formulation minimizes human labor in formulations and simulations. The sliding mode behavior of the model on a frictionless surface and the three link arm stability are demonstrated via simulation. Challenges for application to natural systems are outlined. Copyright © 2014 John Wiley & Sons, Ltd.     

视频中多特征融合人体姿态跟踪

目的 目前已有的人体姿态跟踪算法的跟踪精度仍有待提高,特别是对灵活运动的手臂部位的跟踪。为提高人体姿态的跟踪精度,本文首次提出一种将视觉时空信息与深度学习网络相结合的人体姿态跟踪方法。方法 在人体姿态跟踪过程中,利用视频时间信息计算出人体目标区域的运动信息,使用运动信息对人体部位姿态模型在帧间传递;考虑到基于图像空间特征的方法对形态较为固定的人体部位如躯干和头部能够较好地检测,而对手臂的检测效果较差,构造并训练一种轻量级的深度学习网络,用于生成人体手臂部位的附加候选样本;利用深度学习网络生成手臂特征一致性概率图,与视频空间信息结合计算得到最优部位姿态,并将各部位重组为完整人体姿态跟踪结果。结果 使用两个具有挑战性的人体姿态跟踪数据集VideoPose2.0和YouTubePose对本文算法进行验证,得到的手臂关节点平均跟踪精度分别为81.4%和84.5%,与现有方法相比有明显提高;此外,通过在VideoPose2.0数据集上的实验,验证了本文提出的对下臂附加采样的算法和手臂特征一致性计算的算法能够有效提高人体姿态关节点的跟踪精度。结论 提出的结合时空信息与深度学习网络的人体姿态跟踪方法能够有效提高人体姿态跟踪的精度,特别是对灵活运动的人体姿态下臂关节点的跟踪精度有显著提高。     

Application of a neural network to the generation of a robot arm trajectory

We propose a neural network model generating a robot arm trajectory. The developed neural network model is based on a recurrent-type neural network (RNN) model calculating the proper arm trajectory based on data acquired by evaluation functions of human operations as the training data. A self-learning function has been added to the RNN model. The proposed method is applied to a 2-DOF robot arm, and laboratory experiments were executed to show the effectiveness of the proposed method. Through experiments, it is verified that the proposed model can reproduce the arm trajectory generated by a human. Further, the trajectory of a robot arm is successfully modified to avoid collisions with obstacles by a self-learning function.This work was presented, in part, at the 9th International Symposium on Artificial Life and Robotics, Oita, Japan, January 28–30, 2004     

Analytic generation of workspace using the robot kinematics

When designing workplaces, controls should be placed within the reach of an operator's arm or foot for guaranteeing effective performance. In designing a workplace which must cater to a wide range of operator size, it might be sufficient to plan only for the ‘average person’. Static arm reach measurements which are taken in conventional, standardized positions provide necessary information, but they cannot be applied to dynamic situations directly. To obtain reach envelope or workspace of the human body not by direct measurement but by analytic generation, data on range of joint motion(ROM) are required as an input. The purposes of this research are to measure the range of motion of two degrees of freedom for Korean young males, and to propose an approximate algorithm to generate the workspace of the human body including foot and trunk motion, in which joint mobility of two degrees of freedom motion are considered. The robot kinematics was employed to represent the human body as a multi-link system.     

Improved dynamics model for human standing posture control using functional neuromuscular stimulation

Many studies on the use of functional neuromuscular stimulation for stabilizing the standing posture of paraplegic patients are conducted in computer simulations to minimize possible risks. Here, a five degrees of freedom skeletal-musculotendon muscle activation model of human body dynamics is formulated for the purpose of control effectiveness evaluation. By including the effect of arm rotation about the shoulder plus foot tipping about toe or heel in dynamics models, a more realistic response prediction capability can be achieved compared with previous models. A closed loop control simulation example is presented to demonstrate the significance of adding these features to the human body dynamics model.     

Motion optimization of human body for impulse-based applications

This paper presents the optimal motion generation of the kinematic redundant human body model for impulse-based applications. Inspired by effective and safe motion generation of the human body, we propose a motion generation algorithm considering the external and internal impulses. Firstly, the cost function is defined using the closed form model of external and internal impulses. Then, the self-motion is exploited by using the gradient projection method to generate the optimal motions. Furthermore, the proposed methodology is verified through simulations considering a 4-DOF planar human body model for landing on ground and ball kicking applications and a 3-DOF human arm model for sawing task. It is found that, considering the internal and external impulses for optimization, the optimized posture’s results are well identical to the human body behavior in daily impulse-based motions.     

Estimation of Multijoint Stiffness Using Electromyogram and Artificial Neural Network

The human arm exhibits outstanding manipulability in executing various tasks by taking advantage of its intrinsic compliance, force sensation, and tactile contact clues. By examining human strategy in controlling arm impedance, we may be able to understand underlying human motor control and develop control methods for dexterous robotic manipulation. This paper presents a novel method for estimating multijoint stiffness by using electromyogram (EMG) and an artificial neural network model. The artificial network model developed in this paper relates EMG data and joint motion data to joint stiffness. With the proposed method, the multijoint stiffness of the arm was estimated without complex calculation or specialized apparatus. The feasibility of the proposed method was confirmed through experimental and simulation results.      

Laser-Based Tracking of Human Position and Orientation Using Parametric Shape Modeling

Robots designed to interact socially with people require reliable estimates of human position and motion. Additional pose data such as body orientation may enable a robot to interact more effectively by providing a basis for inferring contextual social information such as people's intentions and relationships. To this end, we have developed a system for simultaneously tracking the position and body orientation of many people, using a network of laser range finders mounted at torso height. An individual particle filter is used to track the position and velocity of each human, and a parametric shape model representing the person's cross-sectional contour is fit to the observed data at each step. We demonstrate the system's tracking accuracy quantitatively in laboratory trials and we present results from a field experiment observing subjects walking through the lobby of a building. The results show that our method can closely track torso and arm movements, even with noisy and incomplete sensor data, and we present examples of social information observable from this orientation and positioning information that may be useful for social robots.