CASIA OpenIR  > 毕业生  > 硕士学位论文
基于微惯性技术的人体运动跟踪关键技术研究
其他题名The key technology about human motion tracking with inertial and magnetic technology
弭鹏
2011-05-27
学位类型工学硕士
中文摘要人体运动跟踪关键技术研究在虚拟现实、人机交互、动画制作、互动娱乐、训练仿真、运动分析等方面有着重要的意义。随着微机电系统(MEMS:Micro-electromechanical Systems)技术的发展,价格低、体积小、重量轻、精度高的新型MEMS传感器不断涌现, 利用新型MEMS传感器技术进行人体运动跟踪将具有非常广阔的应 用前景。 本论文主要围绕基于MEMS技术的人体运动跟踪关键技术展开研究,首先系统地研究了基于微惯性技术的人体运动姿态的跟踪方法及微惯性传感器的标定方法;其次,从步态分析入手,对基于微惯性技术的人体运动位置的跟踪方法进行了研究;最后,集成上述方法和技术,设计并初步实现了一个人体运动捕捉系统。本论文的主要工作有: 第一,提出了一种基于四元数的快速自适应的姿态估计算法。该算法融合三轴陀螺仪、三轴加速计和三轴磁力计来获取实时人体姿态。算法对高斯-牛顿算法和自适应滤波技术进行了有机的结合。首先,基于降价的高斯-牛顿方法的实现过程,提出了一种可以避免传统高斯-牛顿方法在迭代中的矩阵求逆过程的方法,从而使得算法的计算复杂性大大降低。其次,考虑到磁场以及人体运动的加速度对姿态跟踪的影响,提出了一种自适应方法用来减少干扰对姿态准确性的影响。该方法与传统的卡尔曼滤波算法相比,大大降低了算法的复杂性;在周围环境存在磁干扰以及瞬时加速度的情况下,还能够对姿态值进行准确的估计。 第二,提出了一种基于遗传算法的传感器标定方法。该方法能在使用现场对课题组自主研制的高精度位姿跟踪装置CASTracker进行快速准确的标定。该方法充分利用传感器输出模值恒定的原理和不同传感器数据叉乘向量的空间旋转角度,通过遗传算法准确的获取了传感器自身的参数信息和非对准角度。并通过实验证明了算法的有效性。 第三,提出了一种简单的基于步态约束获取人体运动位置的初步方案。本文首先分析了基于微惯性技术获取人体运动位置信息的研究现状,参考国内外相关研究,结合课题组自主研制的高精度位姿跟踪装置CASTracker特点提出了基于步态分析的人体运动位置获取方案,并通过实验验证了该方案的可行性。 第四,设计并初步实现了基于微惯性技术的人体运动捕捉系统。该系统集成了课题组自主研制的高精度位姿跟踪装置CASTracker和上述人体位姿跟踪方法,可以实时捕捉人体的运动状态,并进行实时展示。
英文摘要The key technology of human motion tracking system is very important in the fields of virtual reality, human-computer interaction, animation, interactive entertainment, virtual training and kinematic analysis. With the development of MEMS (Micro-electro mechanical Systems), new type MEMS sensors are available which are qualified of low price, small size, light weight and high precision. There is a very wide application prospects to capture human motion utilizing MEMS sensors. The main content of this paper is related to the key technology of human motion tracking based on MEMS technology. First of all, orientation methods of human motion tracking based on inertial technology as well as sensors calibration methods are systematically studied. Then, started with gait analysis, position methods of human motion tracking based on inertial technology is considered. Finally, a human motion tracking system is designed and implemented integrated with above methods. The main contents are presented as follows: Firstly, an adaptive fast human body orientation tracking algorithm is proposed. This algorithm integrates triaxial gyroscope, triaxial accelerometer and triaxial magnetometer to obtain real-time human orientation, which combines Gauss-Newton method and adaptive filter. First, based on the procedure of reduced Gauss-Newton method, a new method which avoids computing matrix inversion during every orientation update is proposed to reduce computation. Second, with respect to disturbance of magnetic field and acceleration of human motion, an adaptive method is presented to reduce orientation accuracy. Compared with the traditional Kalman filter, this method has a small computation and accurate orientation under the condition of magnetic-field interference and transient acceleration. Secondly, a sensor calibrated procedure based on Genetic Algorithm is presented. This method can quickly calibrate CASTracker (the high precision attitude tracking system which is developed by our research group). Taking full advantage of the constant norm of sensor outputs and space rotation angle of cross product by sensors, the Genetic Algorithm can obtain the accurate sensor paremeters and sensors alignments. Experiments show the validity of this method. Thirdly, a preliminary method to obtain joint position based on gait analysis is put forward. This paper analyzes the current research situation of inertial human motion position tracking, refers to d...
关键词惯性跟踪 滤波算法 人机交互 微惯性技术 Mems 传感器标定 Inertial Tracking Filtering Algorithm Human Computer Interaction Mems Sensor Calibration
语种中文
文献类型学位论文
条目标识符http://ir.ia.ac.cn/handle/173211/7592
专题毕业生_硕士学位论文
推荐引用方式
GB/T 7714
弭鹏. 基于微惯性技术的人体运动跟踪关键技术研究[D]. 中国科学院自动化研究所. 中国科学院研究生院,2011.
条目包含的文件
文件名称/大小 文献类型 版本类型 开放类型 使用许可
CASIA_20082801462908(2751KB) 暂不开放CC BY-NC-SA
个性服务
推荐该条目
保存到收藏夹
查看访问统计
导出为Endnote文件
谷歌学术
谷歌学术中相似的文章
[弭鹏]的文章
百度学术
百度学术中相似的文章
[弭鹏]的文章
必应学术
必应学术中相似的文章
[弭鹏]的文章
相关权益政策
暂无数据
收藏/分享
所有评论 (0)
暂无评论
 

除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。