机器人仿人高速步行的关键机理、控制方法和实现技术

High speed walking is an important indicator of biped robots. This project aims to investigate the main factors which stand in way of the fast walking robots. First, a systemic theory including machine design, motion planning and control will be studied to implement efficient and fast walking with human-like gait. Second, the key technologies and prototype machine will be developed. .The scheming work is of 4 folds. The first one is the compensation method of unbalance moments due to high speed walking. The advantage of this proposal is that a torque control method is studied to achieve this object, which may save much energy then a position based control. Secondly, we want to investigate the method of impact reduction at the moment of foot landing. The landing action yields much higher impact force in fast walking. It may result in the instability contact of the foot, and disturbance to the walking. A series of approaches including machine design, motion planning and control will be done to absorb part of the impact energy, or to translate part of the impact energy into limb motions. The third one is coordination control. Biped robots have complicated structure. When all the limb motions are applied simultaneously, it may result in confliction and can’t achieve the desired performance. A control frame will be developed to guarantee that the complicated system is well organized, and the control operation is simple. Finally, Simulations will be done to verify the effectiveness of the proposed methods. Many details, including the contact process, the time delay of the controller, and the precise behavior of key components, will be studied. The goal of simulation is to implement fast walking over 5km/h with human-like gait. Further, experiments on the key technologies of development of a prototype robot will be conducted..If this project goes well, we can establish a systematic theory on high speed walking robots, and develop a series of key technologies and devices in this field, which is important for the next generation of humanoid robots. ..

步行是高效的人类移动方式，其力学和控制的机理具有很高的研究价值。仿人高速步行是步行机器人研究的重要目标之一，目前机器人步行的速度离人类步行存在明显差距。本申请拟分析制约当前机器人高速步行的主要原因，探究解决这些问题的机理，研究高速步行的控制方法及关键技术。首先，步行中肢体运动的不对称性会导致不平衡转矩，拟研究基于力控制的，上肢及躯干的协调补偿不平衡转矩的机理和控制方法；其次，高速步行中机器人摆动腿的触地冲击会造成足与地面接触不稳定、步态变形等现象。本申请拟从机构学、运动规划和控制等方面综合研究减轻冲击影响的机理和方法；第三，研究可靠性高、操作简单的步行机器人协调控制方法；最后，通过仿真集成和优化仿人高速步行的各项机理和方法，实现5km/h以上仿人步行，并在此基础上研究机器人样机的关键部件和实现技术。项目如果实现期望目标，将为研制的高速步行机器人样机打下全面的基础。

本课题的目标是研究机器人高速步行的机理和控制方法，致力于把被动动态步行应用于复杂的类人机器人步行上，实现具有摆臂、躯干控制等的动态高速两足步行。主要完成的工作有5个方面。首先，研究了在摆动腿着地时刻躯干的姿态对步行效率、稳定性等的影响，并提出了动态步行中控制躯干的方法；其次，研究了一种摆臂控制方法，实现了摆臂与步行的协调。本课题组提出的摆臂控制采用力矩控制的方法，该方法可以用于平衡步行中摆腿力矩；第三，针对动态两足步行的要求，研究了一种以躯干运动为核心，分散化、模块化的步行规划方法，该方法实现了集成躯干运动、腿的摆动、摆臂等运动，形成高度协调的步行；第四，以27自由度类人机器人为原型，重新开发了两足步行机器人仿真系统，实现了带有摆臂、躯干姿态控制、动态的两足步行。仿真实验结果显示，机器人步态流畅，最高对应速度（1.80m高机器人）达到5.14km/h；最后，通过上述理论研究和仿真，在机器人的运动结构（自由度）设计、惯性分布设计和驱动能力设计积累了全面的基础，开展了样机研制工作，设计了一套7自由度液压平面两足机器人系统和一套27自由度三维两足步行机器人，研制了高性能液压关节、高速通信控制系统等关键技术。此外，本项目也完成了项目申请中提出的学术交流和人才培养等任务。总体上看，项目基本完成了原定计划，其中样机研制工作超出预期的工作量，为两足步行机器人的进一步研究建立了坚实的基础。