【作者】 王斌锐；

【导师】 徐心和；

【作者基本信息】 东北大学 ， 模式识别与智能系统， 2005， 博士

【摘要】 双腿行走机器人是一种可模拟人类双腿行走功能的高级智能机器人,容易适应人类生活环境,具有代替人类在危险环境下进行重复、高强度、高精度工作的潜力。目前研制的双腿行走机器人步态和人还存在明显差异。生物医学康复领域研究用微处理器控制的智能假肢来代偿残疾人残缺的肢体。装配智能下肢假肢的残疾人行走步态同正常人很相似。把高级智能假肢——智能仿生腿引入到双腿行走机器人研究中来,必将促进行走机器人的研究。人类正常行走情况下,双腿对称。但在特殊情况下,如下肢残疾、崴脚等情况下,双腿可在不对称情况下行走。不对称双腿行走也是人类智能的体现。研究不对称双腿行走可增强双腿行走机器人行走功能。 本文将智能仿生腿和双腿行走机器人集成起来研究,提出一种新型机器人研究模式——异构双腿行走机器人(Biped Robot with Heterogeneous Legs, BRHL)。其中一条腿采用普通行走机器人模式,称为人工腿;另一条腿采用智能仿生腿。智能仿生腿的加入从关节机构上增强机器人的仿人特性。异构双腿行走机器人为智能仿生腿研究提供开发平台,同时可用于研究人体与智能肢体之间的协调和不对称双腿的行走等问题。 在详细论述异构双腿行走机器人研究意义、内容和方法基础上,本文首先给出了异构双腿行走机器人总体组成。由智能阻尼器控制的4-bar封闭链仿生膝关节转动中心是变化的。通过对膝关节机构参数的多变量最优化设计,保证仿生腿各关节中心点跟踪给定轨迹,且阻尼器能耗小。 采用分割建模方法,分别建立仿生腿和人工腿的运动学模型。采用多体系统动力学建模方法,分别建立双腿在支撑相和摆动相时,带约束和无约束的拉格朗日动力学模型。并从广义变量、广义力、约束之间的关系,分析了动力学模型的求解方法。建立了针对行走的双腿协调动力学模型,并简单分析了双腿之间的耦合。针对假脚弹性形变,给出了假脚建模方法。 将双手协调控制思想引入异构双腿行走机器人中,为双腿协调控制设计了三层递阶控制系统总体结构。用主从式双腿协调控制实现机器人稳定行走的关键是步态规划和仿生腿对人工腿步态一定时差的跟随。由于仿生腿膝关节采用阻尼器控制,不能完全跟踪所有的人工腿步态,所以用多变量SUMT优化方法来求解最优跟踪轨迹。研究采用计算力矩加PD反馈控制方法,控制仿生腿跟踪人工腿摆动相时变步态轨迹。基于Lyapunov方程分析了控制方法收敛性,得出在模型存在误差情况下,理论上无法实现无偏跟踪。为此引入一阶P型开闭环迭代学习控制。开闭环学习控制是学习控制研究的前沿。在证明迭代学习控制收敛性基础上,给出学习增益选择方法。针对完整行走步态,讨论了重复控制的应用。为保证双腿运动协调,研究了基于编码器和6维力/力矩传感器信息的步态在线检测和调整方法。 异构双腿行走机器人机构复杂,研究较困难。利用Pro/E,ADAMS和MATLAB Simulink建立联合仿真平台。在Pro/E中建立了异构双腿行走机器人的三维实体模型。通过Mech/Pro接口将模型导入ADAMS中形成虚拟样机,并在ADAMS中建立虚拟环境。在Simulink中建立各种控制算法的仿真模块。在联合仿真平台上,针对异构双腿行走机器人进行了运动学、动力学仿真。并对计算力矩加PD反馈控制和迭代学习控制进行了控制联合仿真。比较了开闭、环学习控制和单独开或闭环学习控制收敛速度。虚拟样机联合仿真验证,确保控制算法的有更多还原

【Abstract】 Biped robot with two legs is a kind of advanced humanoid robots that can walk like human being and have ability of doing repetitive, hard and exact-need works instead of human being in his living environment. Now biped robot already can walk using two legs, but gait is not like human being’s. In biomedicine rehabilitation field, the intelligent prosthesis controlled by micro-process unit is studied for amputee, and intelligent prosthetic above-knee limb can simulate human leg’s moving and has humanoid knee joint. The gait of amputee with intelligent prosthetic limb is more like health people gait. It can improve biped robot to introduce the study methods and technologies of intelligent prosthesis into biped robot study. Human being’s two legs and their gait are symmetrical in normal walking, but under lame or other knee trauma situation, two legs are heterogeneous. The walking with heterogeneous legs is advanced function of human being and should be studied.Integrating intelligent prosthesis and biped robot study, a new-style robot, biped robot with heterogeneous legs (BRHL) that is composed of a leg of normal biped robot, called artificial leg (AL), and an intelligent bionic leg (IBL), a kind of advanced intelligent prosthesis, is proposed. BRHL provide a good test-bed for study of intelligent bionic leg that has good bionic characteristics, and can be used to study the coordination between human being and intelligent limb, and the walking with heterogeneous legs.Following the meaning, contents and methods of BRHL study, the mechanism and form of BRHL is given. 4-bar closed-chain knee joint controlled by intelligent MR damper has polycentric characteristic of human knee joint. Optimization computation of 14 mechanism parameters of 4-bar knee joint is used to make it track human knee joint and damper cost minimum energy.Using division modeling method, the kinematics and Lagrangian dynamics models of artificial leg and intelligent bionic leg in stance phase and swing phase are built. Constraint conditions are different in stance and swing phase. According to the amount of general variables, general force and constraints, dynamics model’s classification and solving methods are discussed. For sample waling, using crotch point motion as two legs’ task object, the coordination dynamics model is established and coupling of two legs is analyzed. In order to walk like human being, the flexibility of prosthetic foot used for amputee must be considered and flexibility modeling method is studied.The master-slave dual-leg coordination control strategy is pointed, and is compared with dual-arm coordination control. For coordination control, three level hierarchial architecture control system is given. The key problem of master-slaver dual-leg coordination is the gait tracking of IBL to AL in reasonable phase difference. Considering the 4-bar knee joint is controlled by MR damper, IBL cannot follow arbitrary AL gait perfectly. So multi-variable optimization using SUMT algorithm is used to get optimal gait tracking which is a nonlinear, strong coupling, time-varying control problem. Computed torque with PD feedback control is studied and convergence is proved using更多还原