气动肌肉驱动的仿青蛙跳跃机器人及其关键技术的研究

发布时间：2018-05-21 08:55

本文选题：青蛙 + 跳跃机器人　；参考：《哈尔滨工业大学》2016年博士论文

【摘要】：跳跃机器人可以广泛地应用于考古、星际探测、反恐和资源勘探等领域。相对于轮式和行走式机器人,跳跃式机器人的特点是利用跳跃运动前进,具有移动速度快、越障能力强的特点。但是跳跃机器人的运动过程中与地面间的接触情况不断变化,存在着与地面碰撞冲击,使得机器人的动力学过程存在着连续状态和碰撞离散状态相混杂的特点,具有高度的非线性。另外在给定特定任务的情况下,如何合理规划任务空间和关节空间的轨迹,也是一个难点。传统的跳跃机器人多采用“电机+弹簧”的方式为跳跃运动提供动力,这种传动方式采用齿轮、棘轮等复杂的传动机构,容易出现卡死等故障,本文针对这种不足,采用气动人工肌肉作为动力装置,利用气动肌肉的输出力大、传动装置简单的优点,进行跳跃机器人的机构设计。但是气动肌肉具有高度的非线性,力-位移存在滞环特性,力学模型的建立是一个难点。由于力学模型非线性和气体可压缩性特点,精确的位置控制方法的研究也是一个难点。本文在课题组对青蛙的生理结构和跳跃特点已经进行深入研究的基础上,以气动肌肉作为驱动器,构建仿青蛙跳跃机器人本体,对机器人的运动学和动力学特性、气动肌肉的力学模型、实现规则地形下的特定跳跃任务时的轨迹规划、气动肌肉驱动关节控制策略和机器人运动控制器等问题进行研究。本文首先简要分析青蛙的生理结构和跳跃运动特点,建立青蛙跳跃运动等效六杆模型,在ADAMS中对六杆机构进行优化仿真,为机器人机构设计提供参考；对机器人的后肢进行设计,并设计出前肢和整机；根据机器人的状态信息采集的需要,选用相关传感器。分析机机器人跳跃过程的欠驱动特性,将跳跃过程划分为不同的子相,建立统一的运动学模型；对跳跃过程的动力学特性进行分析,基于拉格朗日方程分别建立连续动力学方程和碰撞离散相动力学方程,并分析不同子相运动状态切换的条件。气动人工肌肉的力学特性是机器人设计和控制的基础,因此对气动人工肌肉的力学特性进行研究。采用机理建模和实验建模两种手段对气动肌肉的力学行为进行建模。机理模型以Chou理想模型为基础,考虑橡胶壁弹性、纤维网弹性和内部摩擦等因素的影响；为了实际控制的需要,利用实验手段了建立现象模型；对于PAM内部复杂的充气和排气过程,利用实验数据建立了排气阶段和充气阶段的现象模型。以规则地形下的给定跳跃高度和远度为任务目标,研究轨迹规划问题。对欠驱动关节的求解问题进行深入分析。以地面对机器人反作用力最大值最小为目标,对任务空间的轨迹进行优化；在任务空间轨迹规划的基础上,以消耗的主动力矩最小为目标,对关节空间进行轨迹优化,并进行仿真。设计机器人的运动控制器。对气动肌肉驱动关节的位置控制策略进行研究,构建单自由度的气动肌肉驱动关节实验平台,以实验平台动力学模型和所建立的气动肌肉实验模型为基础,进行PID串级位置控制方法和RBFNN-PID串级位置控制方法的研究。基于气动肌肉驱动关节的位置控制策略,建立机器人的控制器,使用RBFNN-PID串级控制方法对各个关节分别进行控制,并进行matlab/adams联合仿真。设计了机器人的控制系统,并进行位姿调整和跳跃运动的实验研究。以嵌入式微控制器为核心,构建机器人的控制系统。对单条后肢的位姿调整性能和跳跃性能进行实验研究,验证气动肌肉驱动关节控制策略的有效性和轨迹规划方法的正确性；对机器人的位姿调整和跳跃性能进行实验研究,验证机器人控制器的有效性、轨迹规划方案的可行性以及采用气动肌肉作为驱动器构建跳跃机器人的可行性。
[Abstract]:Jumping robots can be widely used in the fields of archaeology, interstellar detection, counter-terrorism and resource exploration. Compared with wheeled and walking robots, jumping robots are characterized by jumping motion, fast moving speed, and strong ability to cross obstacle. But the contact with the ground during the movement of the jumper is not the same. There is a collision with the ground, which makes the dynamic process of the robot have the characteristics of continuous state and collision discrete state, and it has high nonlinearity. In the case of given specific task, it is also a difficult point to plan the task space and the trajectory of joint space reasonably. The traditional jumping robot is also a difficult problem. By using the "motor + Spring" way to provide power for jumping movement, this transmission mode uses complex transmission gear, ratchet and other complex transmission mechanism, easy to die and other faults. In this paper, the pneumatic artificial muscle is used as the power device, the output force of the pneumatic muscle is large and the advantages of the simple transmission device are leaped. The mechanism of the robot is designed. However, the pneumatic muscle has a high nonlinearity, the force displacement has the hysteresis characteristics. The establishment of the mechanical model is a difficult point. The study of the precise position control method is also a difficult point because of the mechanical model nonlinearity and the gas compressibility. On the basis of the in-depth study, a frog jumping robot is constructed with the pneumatic muscle as the driver, the kinematics and dynamics characteristics of the robot, the mechanical model of the pneumatic muscle, the trajectory planning of the specific jumping task under the regular terrain, the control strategy of the pneumatic muscle driving joint and the robot motion control. This paper first briefly analyzes the physiological structure and jumping characteristics of frogs, establishes the equivalent six bar model of frog jumping movement, optimizes the simulation of the six bar mechanism in ADAMS, provides reference for the robot mechanism design, designs the hind limbs of the robot, and designs the forelimb and the whole machine; according to the machine, the machine is designed and the machine is designed. For the needs of the state information collection, the related sensors are selected. The underactuating characteristics of the jumping process of the robot are analyzed, the jumping process is divided into different subphases and a unified kinematic model is established. The dynamic characteristics of the jumping process are analyzed, and the continuous dynamic equation and collision discrete phase are established based on the Lagrangian square path respectively. Dynamic equations and analysis of the conditions for the switching of different subphase motion states. The mechanical characteristics of the pneumatic artificial muscles are the basis of the robot design and control. Therefore, the mechanical properties of the pneumatic artificial muscles are studied. The mechanical behavior of the pneumatic muscles is modeled by two means of mechanism modeling and experimental modeling. The mechanism model is Chou On the basis of the ideal model, the effects of rubber wall elasticity, fiber net elasticity and internal friction are considered, and the phenomenon model is established by experimental means for the needs of actual control. For the complicated aeration and exhaust processes in PAM, the phenomenon model of the exhaust stage and the inflating stage is established by the experimental data. With the given jumping height and distance as the task target, the trajectory planning problem is studied. The problem of solving the underactuated joint is deeply analyzed. In order to minimize the maximum value of the robot counterforce, the trajectory of the task space is optimized. On the basis of the task space trajectory planning, the goal is to minimize the active torque. The joint space is optimized and simulated. The motion controller of the robot is designed. The position control strategy of the pneumatic muscle driven joint is studied. The experimental platform of a single degree of freedom pneumatic muscle driving joint is constructed. Based on the dynamic model of the experimental platform and the established experimental model of the pneumatic muscle, the PID cascade is carried out. The position control method and the RBFNN-PID cascade position control method are studied. Based on the position control strategy of the pneumatic muscle driven joint, the robot controller is established, the RBFNN-PID cascade control method is used to control the joints respectively, and the joint simulation of the matlab/adams is carried out. The control system of the robot is set up and the position and attitude adjustment is set up. The experimental research on the whole and jump movement. The control system of the robot is constructed with the embedded micro controller as the core. The experimental study on the position and posture adjustment performance and jumping performance of the single hind limbs is carried out to verify the validity of the pneumatic muscle driving joint control strategy and the correctness of the trajectory planning method, and the position and posture adjustment and jumping ability of the robot. The experiment can be carried out to verify the effectiveness of the robot controller, the feasibility of the trajectory planning scheme and the feasibility of using the pneumatic muscle as the driver to construct the hopping robot.
【学位授予单位】：哈尔滨工业大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：TP242

【相似文献】