基于动量控制的人型机器人实时稳定全身运动规划  

Momentum control for stable whole body motion of humanoid robots

作　　者：陈毅鸿[1] 陈启军[1] 

机构地区：[1]同济大学电子与信息工程学院,上海201804

出　　处：《控制理论与应用》2014年第6期757-763,共7页Control Theory & Applications

基　　金：上海科委科学基础研究重点资助项目(12JC1408800);国家自然科学基金资助项目(61203344)

摘　　要：本文介绍了一种通过动量控制来生成人型机器人稳定的全身运动生成算法.首先,本文引入了基于浮体坐标系的运动学模型来实现全身运动规划以及工作空间的拓展;并为了解决因此产生的实时运动规划中的稳定性问题,本文定义、分析、求解了机器人运动中的动量因素来作为运动过程中的自平衡器;结合任务空间法同时实现了末端轨迹跟踪和自平衡约束;最后通过仿真和实体机器人实验验证了最终算法的有效性.通过运用本文算法,机器人实现了在屈体向前以及动态跟踪人体运动两个应用场景中的自平衡,并通过对比有无平衡器的质心曲线等数据证明了算法的有效性以及必要性.A momentum control method is proposed to solve the balance problem of a full-body humanoid robot on motion planning. First of all, a kinematic model based on the free-flying coordinate is developed to deal with whole body movements and extend the reachable space of the robot;then, in order to keep balance on the real-time motion planning, momentum parameters of robots are introduced, analyzed and calculated. Finally, the task space method is employed to simultaneously realize end-effectors’ tracking and the auto-balance constraint. Experiments and simulations are carried out on a real full-body humanoid robot to validate the performance of the final policy. When the proposed method is applied to a robot, it succeeds in bending the torso forward and tracking human’s movements by human movement capture device. Comparisons of the results with those obtained without the employment of the proposed method demonstrate the effectiveness and the necessities of this algorithm.

关 键 词：浮体运动学 动量控制 自平衡 人型机器人 实时性 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置] TP273[自动化与计算机技术—控制科学与工程]