基于姿态传感器的人体步态预测算法设计与实现

发布时间：2018-03-21 02:32

本文选题：外骨骼　切入点：步态数据　出处：《西南交通大学》2016年硕士论文　论文类型：学位论文

【摘要】：下肢外骨骼系统是一类人体可穿戴的机器人,通过为人体提供外力支持,达到降低人体的负荷、提高人的持久运动能力的目的,将人的智能和机器人的力量结合在一起,通过机器人完成仅靠人自身无法完成的任务,有广泛的应用前景。外骨骼的执行机构运动到目标位置需要一定的时间,而此过程人体的行走动作已经改变,机械骨骼运动步态滞后于人体运动步态,从而影响穿戴者的行走行为。为了尽量不干扰穿戴者的行走,外骨骼的控制系统的参考信号应当超前于人体的运动状态,因此需要对人体步态数据进行预测。针对慢走、正常行走、慢跑等三种运动模态,设计一套人体步态数据采集系统,采集和预测人体步态数据。首先,以动作捕捉系统OptiTrack采集的人体步态数据为数据源,仿真LMS自适应滤波器、RLS自适应滤波器、奇异谱分析法和非线性时间序列Takens算法,并对四种数据预测算法进行对比分析。仿真结果表明,LMS自适应滤波器和Takens算法更适合实现到硬件中,在线预测人体步态数据。其次,在团队开发的基于ARM的硬件平台上完成步态数据采集系统的软件部分。为该硬件电路编写嵌入式程序,实现电路的通讯、数据处理等功能。用Visual C++编写串口通讯上位机软件。将LMS自适应滤波器和Takens算法实现到硬件平台,在线实时预测人体的步态数据,实验结果表明,Takens算法比LMS自适应滤波器有更好的预测效果,更适合人体步态数据预测。最后,实现外骨骼与步态数据采集系统联调。实验结果表明,本文设计的步态采集系统,能够为外骨骼控制器提供可靠的参考信息,实现了外骨骼跟随人体运动的功能,同时解决了外骨骼步态滞后人体步态的问题。
[Abstract]:Exoskeleton system of lower extremity is a kind of wearable robot. By providing external force support to the human body, it can reduce the load of human body and improve the ability of lasting movement, and combine human intelligence with the strength of robot. The robot can accomplish tasks that cannot be accomplished by human beings alone, which has a wide application prospect. It takes some time for the exoskeleton actuators to move to the target position, and this process has changed the movement of the human body. The mechanical skeleton gait lags behind the human body's walking gait, which affects the wearer's walking behavior. In order to avoid interfering with the wearer's walking, the reference signal of the exoskeleton control system should be ahead of the human body's movement state. Therefore, it is necessary to predict the human gait data. Aiming at the three motion modes of slow walking, normal walking and jogging, a gait data acquisition system is designed to collect and predict the human gait data. Based on human gait data collected by motion capture system (OptiTrack), this paper simulates LMS adaptive filter, singular spectrum analysis method and nonlinear time series Takens algorithm. The simulation results show that LMS-adaptive filter and Takens algorithm are more suitable for hardware and on-line prediction of human gait data. The software of gait data acquisition system is completed on the hardware platform based on ARM developed by the team. The embedded program is written for the hardware circuit to realize the communication of the circuit. Data processing and other functions. Using Visual C to write serial communication host computer software. The LMS adaptive filter and Takens algorithm are implemented to the hardware platform, on-line real-time prediction of human gait data, Experimental results show that Takens algorithm has better prediction effect than LMS adaptive filter, and is more suitable for human gait data prediction. Finally, the combination of exoskeleton and gait data acquisition system is realized. The experimental results show that the gait acquisition system designed in this paper, It can provide reliable reference information for exoskeleton controller, realize the function of exoskeleton following human body motion, and solve the problem that exoskeleton gait lags behind human gait.
【学位授予单位】：西南交通大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TP242

【相似文献】