外骨骼机器人的非结构地面行走步态分类算法

发布时间：2018-03-31 03:29

本文选题：外骨骼　切入点：支持向量机　出处：《机器人》2017年04期

【摘要】：针对人体运动随机性和非结构地面等因素造成在不同类型地面上足底力(GRF)差异大的情况,研发了一套配有鞋底压力传感器、用于实时检测足底力变化的实验靴,提出了基于PSO-SVM(基于粒子群优化算法的支持向量机)的步态分类方法.根据足底受力云图,该实验靴中冗余布置了7枚压力传感器.对人行走在步行机(5 km/h)、水平硬路面和野外草地上的足底力进行了采集和处理.将基本组的足底力作为训练集,预设对应的标签值.基于这些训练集,构建了普通的分类器I和基于粒子群优化算法的支持向量分类器II,并分别利用它们对水平硬路面和野外草地行走的数据进行分类检验.实验结果表明,与普通的分类器I相比,该优化算法不仅对复杂地面上的行走步态分类具有明显的优势,对水平硬路面和野外草地的行走步态识别正确率分别提升了32.9%和42.8%,而且能在某些鞋底传感器发生故障后保持较快的寻优速度和较好的鲁棒性.
[Abstract]:In view of the large difference of foot force GRFs on different types of ground caused by random and unstructured ground, a set of shoe pressure sensors was developed to detect the change of foot force in real time.A gait classification method based on PSO-SVM (support Vector Machine) is proposed.According to the plantar stress cloud map, 7 pressure sensors were arranged in the experimental boots.The foot force of walking on a walking machine (5 km / h), horizontal hard pavement and field grass was collected and processed.The plantar force of the basic group is taken as the training set and the corresponding label value is preset.Based on these training sets, a general classifier I and a support vector classifier II based on particle swarm optimization algorithm are constructed.The experimental results show that compared with the conventional classifier I, the proposed algorithm not only has an obvious advantage in the classification of walking gait on complex ground, but also has obvious advantages in the classification of walking gait on complex ground.The correct recognition rate of walking gait for horizontal hard pavement and field grassland is increased by 32.9% and 42.8%, respectively, and it can keep faster searching speed and better robustness after some soles sensor malfunction.
【作者单位】：中国人民解放军军械工程学院军械技术研究所;东南大学机械工程学院;
【基金】：国家自然科学基金(51575100) 军内科研项目(01060603)
【分类号】：TP242

【相似文献】