基于Kinect下肢康复训练机器人人体步态分析

发布时间：2017-12-28 02:24

本文关键词：基于Kinect下肢康复训练机器人人体步态分析　出处：《沈阳工业大学》2017年硕士论文　论文类型：学位论文

【摘要】：根据统计来看,在2015年就有2.22亿的人口是60岁乃至于60岁以上,达到总人口比例的16.15%;估计到了2020年,将有2.48亿是老年的人口,人口老龄化的水平也将到17.17%,这里有3067万人是年龄较大的80岁以上人口;直到2025年,将突破3亿人是60岁以上的,这将是国家面临着非常严重的老龄化时期。而老年人的人体机能逐渐下降,会导致下肢不协调而跌倒。除外,因疾病、交通、工伤事故等原因造成的下肢病残人员数量也不断增加。由于人口年龄老化加剧和机械损伤也逐年上升,这些伤残人的生活质量下降,同时使其社会和国家经济方面加重负担,抑制了国家经济快速发展。为了提高下肢有障碍的人群的身体机能,需要对他们进行合适的训练。在训练时,获得下肢的步态信息进行分析,这有助于减小意外发生的概率。本课题就基于Kinect传感器搭建一个康复训练机器人,通过Kinect来检测获取人体下肢的步态信息,这里主要是获取人体下肢所贴有八个白色标记点的信息,这些信息数据主要是Kinect水平面到各个标记点的水平距离和各个标记点离地面的垂直高度,并通过这些数据计算出人体下肢膝关节角度。然后通过对膝关节角度和进行时间序列的建模并结合卡尔曼滤波进行预测和估计,同时,通过人体模拟各种步态的实验进行了验证这种方法的合理性和有效性。接着根据得到各类步态实际值和预测值进行差值,运用滑动平均的方法来步态识别,并详细而深入的研究滑动平均法识别步态影响情况,最后用滑动平均法对模拟的具有对称性和非对称性的正常步态的实验进行识别。识别时只需将计算得到的步态序列的预测值与估计值之差进行滑动平均看得到偏差均值的波动范围来判别,在波动允许的范围内说明步态处于正常的,反之,就是异常步态,当异常步态出现时刻同时就会开启报警信号,等待医护人员救援。
[Abstract]:According to statistics, in 2015 222 million of the population is 60 years old and over the age of 60, the total population reached 16.15%; estimated that by 2020, there will be 248 million elderly population, the aging of the population level will be 17.17%, there are 30 million 670 thousand people who are older people over the age of 80; until in 2025, 300 million people over the age of 60 will be a breakthrough, this will be the country facing the aging period is very serious. In the elderly, the function of the human body gradually decreases, which leads to the fall of the lower extremities. In addition, the number of disabled persons in the lower extremities caused by diseases, traffic, industrial injuries and other causes is also increasing. Due to the aging of population and the increase of mechanical damage, the quality of life of these disabled people has been reduced, and their social and national economic burden has been aggravated, which has suppressed the rapid development of the national economy. In order to improve the body function of the lower extremities, it is necessary to train them properly. In training, the gait information of the lower extremities is analyzed, which helps to reduce the probability of accident. This topic on the Kinect sensor to build a rehabilitation training robot based on the detection of lower limb gait information acquisition by Kinect, here is the main access to the lower limb of the human body with eight white marker information, these data are mainly Kinect level to each marked point horizontal distance and vertical height of each marker from the ground, and through these data to calculate the human knee joint angle. Then we predict and estimate the knee angle and time series and combine with Calman filter, and verify the rationality and effectiveness of this method by simulating various gait experiments of human body. Then according to the various gait actual value and predictive value of the difference, using moving average method to gait recognition, gait recognition research of moving average method and effect of detailed and in-depth, finally the simulation of normal gait symmetry and non symmetry by moving average method of experimental identification. Only when the prediction of gait recognition sequence calculated the value and estimated value of the difference of sliding average deviation of the mean fluctuation range to determine, in a normal gait, and that in the range permitted, is abnormal gait, when abnormal gait time at the same time opens the alarm signal, waiting for rescue medical personnel.
【学位授予单位】：沈阳工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP242

【参考文献】