人体运动跟踪传感器设计与标定技术研究
文献类型：学位论文
| 作者 | 李文明 |
| 学位类别 | 工程硕士 |
| 答辩日期 | 2012-05-31 |
| 授予单位 | 中国科学院研究生院 |
| 授予地点 | 中国科学院自动化研究所 |
| 导师 | 杜清秀 |
| 关键词 | 惯性跟踪 虚拟现实 人机交互 人体运动跟踪 姿态融合 inertial tracking virtual reality gesture fusion human computer interaction human motion tracking |
| 其他题名 | Human motion tracking sensor design and study on calibration method |
| 学位专业 | 计算机技术 |
| 中文摘要 | 人体运动跟踪技术是人机交互、虚拟现实领域的重要研究内容之一。随着微机电系统（MEMS：Micro ElectroMechanical System）传感器技术的发展，价格低、体积小、重量轻、精度高的新型大规模集成MEMS传感器不断涌现，为惯性人体运动跟踪技术提供了更成熟的硬件实现平台，使得进行大范围的人体运动跟踪成为可能，因此，利用新型的MEMS传感器进行人体运动跟踪将具有非常广阔的应用前景。 本论文主要围绕人体运动跟踪传感器设计、传感器标定方法与姿态数据融合方法研究及人体运动跟踪系统的设计与实现三部分依次展开。本论文的主要工作如下： 1、结合当前MEMS传感器发展现状，设计并研制了一种小型化、低功耗的人体运动跟踪传感器。首先，该人体运动跟踪传感器采用目前集成度最高的MEMS传感器来设计以尽量减小体积；其次，为确保标定方法和数据融合算法的性能和实时性，该人体运动跟踪传感器采用DSP处理器作为核心处理单元；最后，通过串行SCI总线和CAN总线通信端口分别提供单个人体运动跟踪传感器或者多个人体运动跟踪传感器与PC机的通信。本文所有实验均是基于该硬件平台进行的，并通过相关实验验证了人体运动跟踪传感器平台设计的合理性。 2、提出一种简单、快捷的基于遗传算法的惯性和磁传感器标定方法，并移植了互补滤波姿态融合方法。首先分析了三轴传感器（三轴加速度计、三轴磁力计和三轴陀螺仪）误差来源，并建立了三轴传感器误差模型，主要包括传感器的零漂、比例因子、非正交误差和非对准误差；接着给出目标优化函数，根据采样数据和目标优化函数，采用基因遗传算法对误差参数进行拟合。该方法能在使用现场对人体运动跟踪传感器进行快速准确的标定。并通过实验验证了该方法的有效性。其次，在标定方法基础上设计了人体运动跟踪传感器整体标定流程，并设计了相关标定软件。最后，介绍本文移植的互补滤波姿态融合算法。 3、设计并实现了基于人体运动跟踪传感器的人体运动跟踪系统。该系统通过通过自主设计、研制的人体运动跟踪系统控制器实时获取多个人体运动跟踪传感器数据，实时跟踪人体的运动状态并在PC机上的虚拟人体运动模型上重现。实验结果表明，该系统有抗干扰和跟踪范围广等优点。 |
| 英文摘要 | Human motion tracking technology is the important parts of the fields of the human-computer interaction and virtual reality research. With the development of the MEMS sensor technology, micro-inertial human motion tracking technology, has matured, is a means to an alternative to the optical human motion tracking. Body posture information, by MEMS inertial technology, is not subject to the constraints of the experimental room environment. Clearly, the human motion tracking sensors is the key technologies of MEMS inertial technology for human motion tracking, such sensors have little dependence on the environment, without regard to the occlusion problem of the optical tracking. In theory, it can simultaneously track more than one body's posture information of the same scene. This paper is focused on human motion tracking sensor design, calibration, data fusion and human motion tracking system design, details are as follows: 1. Combined with the current status of the development of MEMS sensors, a human motion tracking sensor hardware design schemes have been proposed. In order to reduce the volume, Highly integrated MEMS sensors was used in the human motion tracking sensors, DSP processor was selected as the core processor to ensure that the run-time integration algorithm, serial communication port SCI and CAN bus communication port respectively to meet the needs of the single motion tracking sensors or more than one motion tracking sensor at the same time. Late experiments also verified the reasonableness of the sensor platform. 2. A MEMS sensor calibration method without precision instruments was proposed. First, three-axis sensor (tri-axis accelerometer tri-axis accelerometer and 3-axis magnetometer)error sources was analyzed, and established a three-axis sensor error model, which including the zero drift, scale factor, non-orthogonal error and non-aligned error, The genetic algorithm is used to solve the error parameter. Sensor self-calibration method was used in the calibration process, such as three-axis gyro error calibrated by tri-axis accelerometer and tri-axis magnetic. Finally the experiments verify the validity of the calibration method. 3. Research-based human motion tracking sensors to obtain the body posture was discussed. For real-time sampling a number of human motion tracking sensors data, the human motion tracking system controller, designed in this paper, can communicate with computer by both the wired and wireless methods. The ex... |
| 语种 | 中文 |
| 其他标识符 | 2009M8014629008 |
| 源URL | [http://ir.ia.ac.cn/handle/173211/7644]  |
| 专题 | 毕业生_硕士学位论文 |
| 推荐引用方式 GB/T 7714 | 李文明. 人体运动跟踪传感器设计与标定技术研究[D]. 中国科学院自动化研究所. 中国科学院研究生院. 2012. |
入库方式： OAI收割
来源：自动化研究所
浏览0
下载0
收藏0
其他版本
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。