节点文献

双足机器人拟人步态规划与稳定性研究

Study on Humanoid Gait Planning and Stability of Biped Robot

【作者】 俞志伟

【导师】 王立权;

【作者基本信息】 哈尔滨工程大学 , 机械电子工程, 2008, 博士

【摘要】 双足机器人具有较高地避障行走能力,能够实现在各种不同复杂地面上行走,在仿人双足机器人技术产业发展中具有广阔的应用前景。研究双足机器人拟人步态,使其实现稳定、快速、拟人行走,具有重要的理论意义和实用价值。本文研究了仿生机构的双足机器人稳定拟人行走步态的关键技术,主要通过研制仿生机构的双足机器人,深入研究双足行走步态稳定性、拟人步态规划与控制方法等。本文分析了人的下肢机构自由度分配及关节驱动方式,设计了具有14自由度仿生机构的双足机器人HEUBR 1。基于人的踝关节运动机理,设计了一种二自由度空间并联机构,应用于双足机器人踝关节设计中。针对串联踝关节运动的不平衡性,优化设计了双足机器人并联踝关节机构,使双足机器人并联踝关节的运动速度和驱动功率峰值均较为平稳,通过在ADAMS中仿真实验,研究优化后并联踝关节的运动性能。分析了零力矩点(ZMP,Zero Moment Point)、压力中心点(COP,Center Of Pressure)的相互关系,得到了在双足机器人与水平地面无粘性力和无吸附力作用下双足机器人的ZMP即为COP的结论。针对拟人步态中支撑面的多种变化,提出了支撑区域描述方法,实现对多点接触时支撑多边形的区域描述;结合ZMP/COP、COG在支撑面内的位置,提出了基于ZMP/COP、COG的综合稳定性判据,通过仿真分析表明比传统的ZMP稳定性判据更能准确地评价步态的稳定性。分析了人的多种行走步态,提出了双足机器人按不同行走速度下的六种步态形式,研究了六种步态的膝关节转角、角速度、驱动力矩、足底支撑面变化及稳定裕度,分析六种步态对应的行走能耗比关系,比较了各种步态的优缺点及适用范围。针对传统的双足机器人采用无足趾屈膝步态行走,提出拟人步态规划方法,在空间三个平面上规划关键步态参数,采用三次样条拟合方法得到对应参数下的连续轨迹。通过调整腿长和足部角度参数可模拟人的多种行走步态,调整关键姿态下踝关节角度参数,能够保证双足机器人稳定行走。分析了传统精确跟踪COG轨迹和ZMP轨迹等步态控制方法,结合人在行走时ZMP平稳移动的特性,提出了一种移动可伸缩倒立摆模型,在约束平面内分析ZMP与COG的运动关系。提出无需精确跟踪期望的ZMP轨迹,但应在有效稳定范围内主动调整ZMP实现对COG轨迹跟踪的控制策略。提出了采用移动可伸缩倒立摆模型中伸缩杆驱动力补偿多杆模型中关节驱动力矩,实现了步态补偿控制。在双足机器人HEUBR 1样机的实验平台上,实现不同步幅下多种步态的稳定行走,比较了三种步态膝关节角度、角速度、驱动力矩、足趾关节转角变化,以及足底压力值、步态的稳定裕度,实验表明了二自由度并联踝关节运动性能的改善,步态的综合稳定性判据的准确性,验证了拟人步态规划与控制方法的可行性。

【Abstract】 Biped robot is more capable of avoid-obstacle walking, and can walk on allkinds of complicated terrains, so it has wide application prospects in the industryof humanoid biped robot technology. There is very important theoreticalsignificance and practical value in the studying the humanoid gait of biped robot,and making it suitable of humanoid walking satbly and fast.This paper mainly study the key technology in the gait of humanoid walkingof humanoid biped robot. And this dissertation makes the study through theresearch of bionic mechanism of biped robot, the stablity of biped waiking gait,the planning of humanoid gait and the control method of it.This paper makes analysis of the freedom distribution and the driving modeof joint in human’s low limbs, and has disigned the biped robot HEUBR1 withbionic mechanism of 14 DOF. Based on the mechanism of human’s ankle jointmovement, this paper disigned a new kind of parallel mechanism with 2 DOF, andapplied it in the design of joint in the biped robot. Considering the imbalancecharacter of series ankle motion, this dissertation optimized the design withparallel ankle joint mechanism in the biped robot, and made the motion velocityand the peak value of driving powerot of the parallel ankle joint mechanism in thebiped robot more stable. Through the analysis of simulation experiment in theADAMS, this paper studied the motion performance of the optimized parallelankle joint mechanism.This paper analyzes the relationship between Zero Moment Point(ZMP) andCenter Of Pressure(COP), and makes the conclusion that the ZMP of biped robotis the COP of biped robot under the condition that there is no viscous force andadsorption force between the biped robot and horizontal ground. According to thedifferences of the supporting surface in the gait of humanoid walking, this paperproposes description method of supporting surface, fulfills the region descriptionsof support polygon under the condition of multipoint contact; based on thecomprehensive consideration of the position of ZMP/COP and COG in thesupporting surface, this paper proposes criterion for comprehensive stabilitywhich is based on the consideration of ZMP/COP and COG, and the result ofsimulation experiment shows that this method is better than traditional criterionfor comprehensive stability in the evaluation of the gait’s stability. This paper analyzes several kinds of human’s walking gaits, then proposesthe 6 different kinds of biped robot’s gait under different walking velocity, andanalyzes the rotation angle of joint, angular velocity, driving torque, the change ofsupporting surface footplate and the stability margin under the 6 different gaits.And this paper analyzes the corresponding energy consumption ratio of waikingof different walking gait, then compares scope of application, advantages anddisadvantages of different gaits. Comparing to the non-toe genuflex gait oftraditional biped robot, this paper proposes the planning method of humanoid gait,and plans the key parameter of gait in 3 dimensions, then uses the cubic splinefitting method to abtain the continuous path corresponding to certain parameters.By adjusting the length of leg and the angle of foot, it can simulate several kindsof human’s walking gait. By adjusting the angle of ankle joint under key attitude,it can make sure that the biped robot walk stably.This paper analyzes the traditional methods of gait control, such as trackingthe COG trajectory and tracking the ZMP trajectory, then compares the stablewalking character of human’s walking. Based on these reseaches, this paperproposes an inverted pendulum model with the charater of mobile and scalable,and analyzes the relationship between the movement of ZMP and COG inconstraint plane. Besides, this paper proposes a new control strategy which needsno accurate tracking of the expected ZMP trajectory but should actively adjust theZMP to track the COG trajectory in effective stable range. This paper presents anew method to fulfill the control of gait compensation by using the driving forceof telescopic handle in the inverted pendulum model with the charater of mobileand scalable to compensate the driving torque of joint in the multi-bar model.Based on the experimental platform of biped robot HEUBR1, this paperimplements stable walking of different gait under different stride frequency, thecompares the angle of knee joint, angular velocity, driving torque, the change oftoe joint’s angle, plantar pressure and stability margin of the gait of three differentgaits. The results show that there is big improvement in the walking of biped robotwhen using the parallel ankle joint with 2 DOF, and more accurate criterion forthe comprehensive stability of different gait, then prove feasibility of planning forhumanoid gait and control method.

节点文献中: 

本文链接的文献网络图示:

本文的引文网络