智能上肢可控变阻抗柔性关节控制与安全路径规划

发布时间：2018-04-23 09:41

本文选题：智能上肢 + 差动绳驱关节　；参考：《沈阳工业大学》2017年硕士论文

【摘要】：智能上肢通常工作于人机交互的环境中,以安全性为首要要求,同时应满足一定的控制精度和响应速度。由于其前端输入指令多为离散的相对模糊的指令而非准确的给定运动路径,智能上肢控制器应具有一定的自主安全路径规划能力,这也是其“智能”的体现。因此,智能上肢设计控制的目标是在较为复杂的日常环境中,在“向左”、“向右”、“抓取”等离散指令下能够安全、准确、迅速且智能地执行使用者的运动意图。文章中智能上肢为自主研发制作的新型6自由度轻量化智能上肢。首先介绍了在设计制作方面进行的多种创新,提出了一种差动绳驱关节,结合轻质铝合金加工的结构以满足智能上肢对于重量、负载能力以及响应速度等各方面的要求。进一步的,根据其关节构型,使用改进D-H方法建立了智能上肢数学模型。并针对关节驱动的耦合特性,对于包含驱动器空间-关节空间-笛卡尔空间之间相互转换的刚体运动学与动力学模型进行了构建并提出一种针对其特殊关节构型的运动学逆解算法。此部分内容是可控变阻抗柔性关节(CIVFJ)的硬件基础,运动模型的建立是可控变阻抗柔性关节控制及安全路径规划的理论基础。在此基础上针对智能上肢可控变阻抗柔性关节控制进行了探讨。本文实现了一种虚拟可控柔性关节并构建其动力学模型,对柔性关节领域广泛使用的控制方法做了较为深入的讨论与比较,并最终提出基于Sigmoid函数的增益自调节控制方法,以实现可控变阻抗柔性关节能在阻抗系数与负载变化的情况下保持理想的位置精度和动态响应特性。这部分研究保证了智能上肢的精度与响应速度,同时为以变阻抗柔性关节为基础的被动规划方法打下基础。之后对于安全路径规划的研究,分为主动、被动与主被动结合三种方式。其中,主动安全路径规划以快速扩展随机树(RRT)方法为基础,采用了一种改进RRT*方法,在有视觉监督的环境下,实现障碍环境中的自主避障。无危险系数判断的被动路径规划方法则在视觉监督缺失或不可靠的环境中,通过关节柔性的改变,保证运动过程中始终保持较低的潜在伤害。最后本文提出的笛卡尔空间中主被动结合的规划方法,能够提升智能上肢在动态复杂环境中的安全运行能力,体现出了智能上肢的安全与智能。各章节中,在Matlab Robotic Toolbox等仿真环境下,对于智能上肢运动学、动力学模型,新提出的逆解算法、变阻抗柔性关节控制方法以及主被动安全路径规划方法进行了仿真,并以此为基础在智能上肢实物上进行了实验,验证了其运动模型的合理性以及提出方法的有效性。
[Abstract]:Intelligent upper limbs usually work in the environment of human-computer interaction. Safety is the most important requirement, and the control precision and response speed should be satisfied at the same time. Since most of the front-end input instructions are discrete relatively fuzzy instructions rather than accurate given motion paths, the intelligent upper limb controller should have a certain ability of autonomous safe path planning, which is also the embodiment of its "intelligence". Therefore, the aim of intelligent upper limb design control is to carry out the user's motion intention safely, accurately, quickly and intelligently under discrete instructions such as "left", "right" and "grab" in complex daily environment. In this paper, the intelligent upper limb is a new 6-DOF lightweight intelligent upper limb. In this paper, a variety of innovations in design and manufacture are introduced, and a kind of differential rope drive joint is proposed, which combines the structure of light aluminum alloy machining to meet the requirements of intelligent upper limb in terms of weight, load capacity and response speed. Furthermore, according to its joint configuration, an improved D-H method is used to establish a mathematical model of intelligent upper limb. And aiming at the coupling characteristics of joint drive, The kinematics and dynamics model of rigid body including the transformation between actuator space joint space and Cartesian space is constructed and an inverse kinematics algorithm for its special joint configuration is proposed. This part is the hardware foundation of controllable variable impedance flexible joint (CIVFJ), and the establishment of motion model is the theoretical basis of controllable variable impedance flexible joint control and safe path planning. On this basis, the controllable variable impedance flexible joint control of intelligent upper limb is discussed. In this paper, a virtual controllable flexible joint is implemented and its dynamic model is constructed. The widely used control methods in flexible joint field are discussed and compared in depth. Finally, a gain self-adjusting control method based on Sigmoid function is proposed. In order to realize the controllable variable impedance flexible joint can maintain the ideal position precision and dynamic response characteristic under the condition of the impedance coefficient and the load change. This part of the research ensures the precision and response speed of intelligent upper limb and lays the foundation for passive programming method based on variable impedance flexible joint. After that, the research on safe path planning is divided into three ways: active, passive and passive. The active secure path planning is based on the fast extended random tree (RRT) method, and an improved RRT * method is adopted to realize the autonomous obstacle avoidance in the obstacle environment with visual supervision. In the absence of visual supervision or reliability, the passive path planning method with no risk coefficient can ensure the low potential injury during motion by changing the flexibility of joint. Finally, the planning method of combining active and passive in Cartesian space is proposed in this paper, which can improve the safety operation ability of intelligent upper limb in dynamic and complex environment, and reflect the safety and intelligence of intelligent upper limb. In each chapter, the kinematics and dynamics model of intelligent upper limb, the new inverse solution algorithm, the variable impedance flexible joint control method and the active and passive safe path planning method are simulated under the Matlab Robotic Toolbox simulation environment. On the basis of the experiments, the rationality of the motion model and the validity of the proposed method are verified by experiments on the intelligent upper limb.
【学位授予单位】：沈阳工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP241

【参考文献】