文章基本信息標題:An Adaptive Method for Gait Event Detection of Gait Rehabilitation Robots期刊:Frontiers in Neurorobotics作者:Jing Ye, Hongde Wu, Lishan Wu, Jianjun Long, Yuling Zhang, Gong Chen, Chunbao Wang, Xun Luo, Qinghua Hou and Yi Xu作者簡介:作者分別來自深圳邁步機器人公司,深圳老年病研究所,中山大學附屬醫院康復科,哈佛醫學院卒中康復實驗室。金句分享IntroductionIn the field of gait rehabilitation robots, synchronization between motion of robots and actual human gait requires that the robot can accurately identify the current gait event and then automatically adjusts its gait phase to achieve human-robot synchronization.(介紹背景)The above researches are most focused on information analysis from sensors mounted on the shank or foot. However, we find that, compared to the thigh, the motion of the shank and foot have higher speed and variable acceleration while walking, which may cause higher accumulated error when calculation the shank or foot angle based on angular velocity detected by IMU.(總結前人工作不足)The focus of this paper is to build a linear regression model, which can detect an HS gait event for different subjects walking at various stride frequencies. Besides, it obtains a set of observable fixed thresholds for TO gait event detection.(文章主要方法及目的)MethodThe conventional thresholding method for gait event detection is based on fixed thresholds. The proposed method adopts adaptive thresholds for HS detection and observable thresholds for TO detections.(方法創新)For the purpose of the present study, only the outputs of the gyroscope orthogonal to the sagittal plane and the forward accelerometer orthogonal to the coronal plane were retained for analysis.(定義主要輸入信號)For HS detection, the data from 20 healthy subjects were classified into training and testing sets for linear regression models, and five different train:test subject ratios were adopted, ranging from 9:1 to 5:5.(檢測模型訓練)ResultThe proposed method can detect all HS/TO gait events precisely. However, the conventional thresholding method misses HS gait event detections when a set of fixed thresholds is used.(主要結論)DiscussionFor HS, because stride lengths affect thigh angle spikes, when a subject wearing the BEAR-H1 walks at one kind of stride frequency, thigh angle may change within a relatively large range. In this case, the model can still detect HS gait events within 3% gait cycle errors.(方法的優越性)Foot pressure sensors used here are considered as a convenient way to generate the labeled data sheet for HS and TO gait event detection. The accuracy of gait detection based on foot pressure sensors may be questioned.(方法可能存在的問題)文章簡介
因此,自動情緒識別技術是諸多領域的剛需,如遊戲娛樂、安保執法、購物、人機互動等。有了它,機器人將能更好地與人類產生交流。一個解決方案在這篇名為《Identifying Emotions from Walking Using Affective and Deep Features》的論文中,研究團隊提出了一種全新的自動情緒識別方法,可以將視頻中行走的人類進行歸類為快樂、悲傷、憤怒或中立 4 種情感類別。