【作者】 王超；

【导师】 卢俊国；

【作者基本信息】 上海交通大学 ， 控制科学与工程， 2015， 硕士

【摘要】 随着时代的发展,机器人渐渐渗透到人们日常生活中。但由于刚性结构的限制,传统的刚性机器人在非结构化环境中的表现难以令人满意。在这样的背景下,作为对刚性机器人的重要补充,软体机器人开始出现在人们的视野里。软体机器人的灵感及设计理念来源于生物界中的软体组织,研究者期望通过对软体组织材料、结构以及运动机理的模仿,制造出一种像哺乳动物舌头、章鱼触手一样具有良好灵活性与延展性的软体机器人。与传统的刚性机器人相比,软体机器人的优势主要表现在材料与结构方面。软体机器人由柔性材料制作,而且内部不包含刚性结构,驱动机构也采用的具有良好柔性的装置。因而软体机器人能表现出良好的灵活性与安全性,从而能在非结构化的环境中自如运动。虽然软体机器人有着传统刚性机器人无法比拟的优势,但材料的柔性使得软体机器人的运动机理变得复杂,这给软体机器人动力学建模、控制器设计等理论分析带来极大的困难,因而严重制约了软体机器人的实际应用。为解决软体机器人的上述问题,本文以线驱动硅胶软体机械臂为研究对象,针对软体机械臂的运动学、动力学建模与控制、形状检测等问题展开一系列的研究,研究内容及创新点概述如下:分析了线驱动软体机械臂的运动特点以及狭小环境对软体机械臂运动的影响,在分段常曲率假设的基础上,建立了线驱动软体机械臂的运动学模型;同时,针对软体机械臂手眼系统,提出了一种能在线估计受限运动学模型中未知参数的自适应控制算法,运用巴巴拉引理,证明了控制算法的稳定性;最后,设计了受限环境下的视觉伺服对比实验,验证了控制算法,并分析了控制参数以及环境对软体机械臂运动的影响。随后,结合几何精确Cosserat梁理论,改进了基于分段常曲率假设的运动学模型,提出了新的运动学模型;在新运动学的基础上,运用凯恩方法,建立了描述软体机械臂三维运动的动力学方程;在动力学建模中,基于集中参数假设,分析了软体机械臂各单元的惯性力与重力;结合粘弹性材料的开尔文模型,分析了各单元的弹性内力;同时研究驱动线的驱动原理,分析了驱动线的驱动力;对动力学方程进行化简与整理,将方程改写为便于数值求解的形式;最后,进行了仿真和实验,并与商业软件RecurDyn结果以及实验结果对比,验证了模型的准确性。最后,简要分析了软体机械臂形状反馈在实际应用中的必要性,随后在光纤布拉格光栅传感器的基础上,设计了一套分布式光纤传感网络以及相应的曲率-挠率测量算法,能将光栅测量到的应变信息换算为光栅处的曲率和挠率;结合光顺曲线理论与三次B样条理论,提出了一种形状检测算法,该算法能根据测量到的曲率、挠率信息计算出软体机械臂中心线的形状;最后,设计了仿真验证了算法的有效性。更多还原

【Abstract】 With the development of technology, robots are utilized in increasingly wider applied fields, which are not confined to traditional industry or scientific research, but part of our daily lives.However, limited by the rigid structure and material, traditional rigid robotic can’t meet the requirements for the specified application or intended use when applied to unstructured environment. Under such circumstances, a kind of novel robot—soft robot appears as an important complement to traditional rigid robot. Compared with traditional rigid robot, soft robot’s advantages are the softness and flexibility due to the soft materials and soft structures,which make it possible for the soft robot to work in the unstructured environment freely.Despite the overwhelming superiority of soft robot,the softness complicates the movement rule of the soft robot,which makes it rather difficult to build the accurate model and design control algorithm, hence limit the applications of soft robot in our daily lives. In order to deal with the problems of soft robot mentioned above, this work focus on not only the kinematics and dynamics of cable-driven soft manipulator, but also the controller and shape detecting algorithm development. The main contents and innovations of this thesis are summarized as follows:Firstly, the kinetic characteristics of cable-driven soft manipulator in narrow and disorder environment are analysed.And the forward kinematics and differential kinematics of cable-driven soft manipulator based on piecewise constant curvature assumption are proposed. In addition, an adaptive control algorithm for the eye-in-hand system of soft manipulator is developed. The adaptive controller enable the convergence of soft manipulator’s motion in narrow environment by estimating the unknown parameters of the differential kinematics online. The contrast experiments are designed to validate the adaptive control algorithm and analyse the impact of related parameters.Then, the forward kinematics are improved based on piecewise constant curvature assumption and a new forward kinematics of soft manipulator is developed by combining the geometrical accurate Cosserat rod theory. In addition, a new three dimensional dynamics of cable-driven soft manipulator based on new kinematics and Kelvin model of viscous-elastic material is proposed, by utilizing the Kane’s method. The dynamics contains four kinds of forces and moments: the inertia forces and moments of each segment of soft manipulator based on lumped parameter assumption; the elastic forces and moments based on Cosserart rod theory and Kelvin model; driven forces and moments based on cable-driven principle. The contrast experiments between numerical simulations and business software(Recur Dyn) are designed to validate the dynmics.Finally, a distributed sensor network based on FBGs and relevant algorithm are proposed to deal with the shape detecting problem of soft manipulator in closed environment. The sensor network can measure the curvatures and torsions of soft manipulator simultaneously. Based on fair curve theory, a shape detecting algorithm, which can translate the curvature and torsion measured by sensor network into the soft manipulator’s shape, is proposed. Eventually, high accuracy for three dimensional shape the shape detecting method introduced in this paper can achieve is validated by three dimensional simulations.更多还原