Human Gait Behavior Interpretation by a Mobile Home Healthcare Robot
Human Gait Behavior Interpretation by a Mobile Home Healthcare Robot
复制标题
移动家庭医疗机器人解释人类步态行为
DOI：
10.1142/s0219519412400210
复制
发表时间：
2012
影响因子：
0.8
通讯作者：
Yoshida Yuki and Yu Wenwei
中科院分区：
文献类型：
--
作者：
Nergui Myagmarbayar;Yoshida Yuki and Yu Wenwei
本研究的最终目标是开发自主移动的家庭医疗机器人，密切监测和评估患者的运动功能，以及他们在家里的训练治疗过程，在紧急情况下提供自动呼叫医疗人员。该机器人的开发将为大多数运动功能障碍患者（MIPs）带来经济、安全和方便的家庭康复，同时减轻治疗师在规范康复方面的巨大负担。为了实现这一目标，我们已经开发了以下程序/算法，用于监测人类活动和识别人类行为：（1）控制程序，用于移动的机器人跟踪和跟随人类主体的三个不同的观点;（2）算法，用于分析下肢关节角度从RGB-D图像从Kinect传感器设置在移动的机器人;和（3）算法，用于识别人类步态行为。在（1）中，开发了侧视点、前/后视点和中间角视点（在两个前视点之间）跟踪。针对移动的机器人在跟踪人体时，由于深度图像混叠和帧间跳跃而导致的骨架点提取误差，采用彩色标记进行深度图像补偿。在（3）中，我们提出了一种基于隐马尔可夫模型（HMM）的人类行为识别，利用下肢关节角度和躯干角度。实验结果表明，与运动跟踪系统相比，关节轨迹可以以高精度测量和分析，并且可以从关节轨迹识别人类行为。
The ultimate goal of this study is to develop autonomous mobile home healthcare robots which closely monitor and evaluate the patients' motor function, and their at-home training therapy process, providing automatically calling for medical personnel in emergency situations. The robots to be developed will bring about cost-effective, safe and easier at-home rehabilitation to most motor-function impaired patients (MIPs), and meanwhile, relieve therapists from great burden in canonical rehabilitation. In order to achieve this goal, we have developed the following programs/algorithms for monitoring human activities and recognizing human behaviors: (1) control programs for a mobile robot to track and follow a human subject by three different viewpoints; (2) algorithms for analyzing lower limb joint angles from RGB-D images from a Kinect sensor setup at a mobile robot; and (3) algorithms for recognizing human gait behavior. In (1), side viewpoint, front/back viewpoint and a middle angle viewpoint (between two former viewpoints) tracking were developed. In (2), depth image compensation with colored markers was implemented to deal with the skeleton point extraction error caused by mixing-up and frame flying of depth image during tracking and following human subjects by the mobile robot. In (3), we have proposed a hidden Markov model (HMM) based human behavior recognition using lower limb joint angles and trunk angle. Experimental results showed that joint trajectory could be measured and analyzed with high accuracy compared to a motion tracking system, and human behavior could be recognized from the joint trajectory.