人体步行捕捉下的双足机器人跟随步行与实验

发布时间：2018-03-15 01:14

本文选题：仿人双足步行机器人　切入点：跟随步行　出处：《哈尔滨工业大学学报》2017年01期 　论文类型：期刊论文

【摘要】：为深入研究服务型仿人机器人实时跟随人步行的问题,提出了基于人体步行运动捕捉的双足机器人步行样本生成方法,并进行了机器人-人跟随步行实验.对PS三维运动捕捉系统在线获取的人体步行样本进行运动学匹配并考虑机器人关节极限约束条件后,得到机器人步行样本,构建机器人仿人步行的样本库;根据笛卡尔空间和关节空间内的运动参数定义机器人与人的步行相似度综合评价,提出基于相似度评价的在线样本检索方法,以"关节角距离"为评价选取拼接点,实现样本在线拼接的样本过渡方法,解决了机器人跟随人进行变速步行的问题;完成了双足机器人跟随不同人进行稳定步行的实验,跟随过程中的距离误差不超过±52 mm,跟随结束后的位置误差不超过±10 mm.
[Abstract]:In order to deeply study the problem of real-time following human walking of service humanoid robot, a walking sample generation method for biped robot based on human walking motion capture is proposed. The robot human follow walking experiment was carried out. The robot walking sample was obtained by kinematics matching of human walking samples obtained online by PS 3D motion capture system and taking into account the limit constraint condition of robot joint. Based on the motion parameters in Cartesian space and joint space, the comprehensive evaluation of walking similarity between robot and human is defined, and an online sample retrieval method based on similarity evaluation is proposed. Taking "joint angle distance" as the evaluation method, the method of sample transition is realized, which solves the problem of robot walking with variable speed, and accomplishes the experiment of biped robot following different people to walk stably. The distance error in the following process is not more than 卤52 mm, and the position error after the following is less than 卤10 mm.
【作者单位】：哈尔滨工业大学机电工程学院;
【基金】：国家高技术研究发展计划(2012AA041601)
【分类号】：TP242

【相似文献】