【作者】 赵正华；

【导师】 张连东；

【作者基本信息】 大连交通大学 ， 机械制造及其自动化， 2011， 硕士

【摘要】 模块化双足机器人在快速行走和越障过程中都有着无可比拟的优越性。根据运动学参数和动力学行为的要求,可以采用标准的连杆模块和关节模块快速装配成满足用户需求的机器人构型。它具有环境适应能力强、维护简单、可实现清洁生产、成本低廉、装配方便等优点。本文研究的模块化双足机器人采用德国工业自动化公司生产的PowerCube模块组装而成,腿部具有12个自由度,上身采用类人形质量块代替,不具备自由度。本文首先对模块化双足机器人的概况进行介绍,通过D-H表示法研究了机器人正运动学并求解出其运动变换矩阵,基于拉格朗日方程推导出机器人的动力学方程,为后期的运动仿真和动态平衡控制提供理论依据。在对F/T传感器仿真研究过程中,通过对传感器本体模型的合理简化,将测力模块简化成弹簧-阻尼-质量块系统,把测力矩模块简化成反力矩测试装置,进行数学建模分析。采用线性二次最优控制理论对传感器控制系统进行建模并在Matlab软件simulink模块中进行仿真。应用PID校正方法来校正控制器系统的响应特性,最后以二阶倒立摆和车作为控制输入,对传感器控制系统进行仿真验证。在对机器人动态平衡研究过程中,提出在单脚支撑期和双脚支撑期基于李群理论利用传感器检测信息实时检测机器人零力矩点(ZMP)的方法,并在实验研究中验证了检测方法的准确性。建立机器人参考坐标系,依据ZMP和质心(COM)约束条件,建立凸优化方程,利用matlab软件MOSEK工具箱进行凸优化编程运算,进而实现COM和ZMP循迹误差、髋关节、膝关节、踝关节关节轨迹规划研究,验证凸优化二阶锥优化算法的有效性和机器人动态平衡行走的可行性。更多还原

【Abstract】 When walking fast or crossing obstacle, modular biped robot has several unparalleled advantages. It can adjust itself to different conditions. It is simply maintained, friendly to environment, cost-effective and easy to be assembled. According to kinematic and dynamic requirements, robot can be assembled by standard link and joint modules. Modular biped robot in this paper is assembled by PowerCube modules which are produced by Schunk company. It has twelve degrees of freedom for the legs and the upper body is simplified with no degree of freedom.Firstly, we introduce the profile of modular biped robot. The forward kinematics of robot using D-H parameters and tansformation matrix are obtained. The dynamic equation of the robot is also attained based on Lagrange method which provides theoretical basis for further simulation.For the F/T sensor simulation, mathematical modeling analysis is done by simplifying force module for spring-damper-mass system and moment module for countertorque detecting device. Then we establish Linear Quadratic optimal controller and simulate it in simulink module of matlab. The response of control system is corrected by PID. The simulation of sensor control system with second order inverted pendulum and cart as input is implemented and the simulation results show that the control method is correct.When studying dynamic balance of the robot, a ZMP formula using Lie Group is given for detecting ZMP of robot in real time. This formula uses information from F/T sensor both in single support phase and in double support phase. ZMP point detecting experiment is carried out and detecting method is verified.At last, convex optimization for balance control is studied. The ZMP and COM constraints are established. The trajectories of hip, knee and ankle joints are obtained using Mosek toolbox in Matlab. The simulation results show that Second Order Cone Convex Optimization is effective and stable walking of robot can be realized.更多还原