绳驱超冗余机械臂通常由很多关节组成，体型纤细，运动灵活性强，非常适合在狭小、非结构化的环境中进行作业。在航天、核电、医疗等领域中常常需要使用绳驱超冗余机械臂帮助人们完成操作任务。然而由于绳驱超冗余机械臂运动时容易受到重力和摩擦等因素的影响，导致运动学建模存在困难，进而影响臂形估计的准确性、轨迹规划算法的可靠性。为了让绳驱超冗余机械臂能够更好的发挥其在复杂环境下的运动能力，本文针对绳驱超冗余机械臂的运动学建模、臂形估计算法、轨迹规划算法展开研究，并通过仿真实验和实物实验验证建模的准确性和算法的有效性。运动学建模方面，研究了由于重力和摩擦等因素导致的联动误差对于绳驱超冗余机械臂运动学建模的影响，并通过实验数据得出联动误差因子的表达形式。而后在考虑了联动误差因子的基础上，采用指数积公式进行运动学建模，包括关节空间到构型空间正运动学建模、关节空间到绳长空间正运动学建模、以及逆运动学建模。运动学建模为后续的臂形估计和轨迹规划算法提供了模型基础。臂形估计方面，提出一种基于左不变卡尔曼滤波算法的多传感器融合的臂形估计算法。首先介绍了左不变卡尔曼滤波算法的优点，对于非线性的复杂系统，左不变卡尔曼滤波算法更稳定。然后分别介绍了双目相机的位姿估计方法及惯性测量单元的动力学模型，并基于左不变卡尔曼滤波算法将双目相机的位姿估计信息、惯性测量单元的信息以及绳长信息进行融合。最后在仿真环境中进行了臂形估计实验，验证了融合算法的有效性。避障轨迹规划方面，考虑到绳驱超冗余机械臂关节约束多导致路径优化的成本函数中具有大量的非线性约束的问题，提出一种基于路径积分的轨迹规划算法，避免对成本函数中不可微分量求导的过程。对于障碍物环境的描述采用更为通用的ESDF方式，便于机械臂在各种环境中的避障规划。在仿真环境中搭建了障碍物的ESDF地图，验证避障轨迹规划算法生成路径的收敛性和平滑性。最后为了验证本文提出的绳驱超冗余机械臂运动学建模的准确性、臂形估计及轨迹规划算法的有效性，自主开发了基于ROS的绳驱超冗余机械臂控制交互平台，并搭建了吊丝系统、动作捕捉系统等硬件设施用于实验。进行了臂形估计实验和避障轨迹规划实验，验证算法在实物上的有效性。

Cable-driven super-redundant manipulators are usually composed of many joints, with slim body size and high flexibility of movement, which are ideal for working in small and unstructured environments. Cable-driven super-redundant manipulators are often used in aerospace, nuclear, medical and other fields to help people complete operational tasks. However, the kinematic modeling is difficult because the cable-driven super-redundant manipulator is easily affected by gravity and friction, which affects the accuracy of manipulator shape estimation and the reliability of trajectory planning algorithms. In order to make the cable-driven super-redundant manipulators perform better in complex environments, this thesis investigates the kinematic modeling, manipulator estimation algorithm, and trajectory planning algorithm of the cable-driven super-redundant manipulator, and verifies the accuracy of the modeling and the effectiveness of the algorithm through simulation and physical experiments.In terms of kinematic modeling, the effect of linkage error caused by gravity and friction on the kinematic modeling of the cable-driven super-redundant manipulator is studied, and the expression of the linkage error factor is derived from the experimental data. After considering the linkage error factor, the exponential product formulation is used for kinematic modeling, including positive kinematic modeling from joint space to configuration space, positive kinematic modeling from joint space to rope length space, and inverse kinematic modeling. The kinematic modeling provides the model basis for the subsequent manipulator shape estimation and trajectory planning algorithms.For manipulator shape estimation, a multi-sensor fusion manipulator shape estimation algorithm based on the left-invariant kalman filter algorithm is proposed. Firstly, the advantages of the left-invariant kalman filter algorithm are introduced, and the left-invariant kalman filter algorithm is more stable for nonlinear and complex systems. Then the positional estimation method of binocular camera and the dynamics model of inertial measurement unit are introduced respectively. Finally, the left-invariant kalman filtering algorithm is used to fuse the position estimation information of the binocular camera, the information of the inertial measurement unit, and the cable length information. The manipulator shape estimation experiments are conducted in the simulation environment to verify the effectiveness of the fusion algorithm.For obstacle avoidance trajectory planning, considering the problem that many joint constraints of the cable-driven super-redundant manipulator lead to a large number of nonlinear constraints in the cost function of path optimization, a path-integration-based trajectory planning algorithm is proposed to avoid the process of derivation of non-differentiable components in the cost function. A more general ESDF approach is adopted for the description of the obstacle environment, which facilitates the obstacle avoidance planning of the manipulator in various environments. The ESDF map of obstacles is built in the simulation environment to verify the convergence and smoothness of the path generated by the obstacle avoidance trajectory planning algorithm.Finally, in order to verify the accuracy of the kinematic modeling of the cable-driven super-redundant manipulator proposed in this thesis and the effectiveness of the manipulator shape estimation and trajectory planning algorithms, a ROS-based cable-driven super-redundant manipulator control interaction platform was developed independently, and hardware facilities such as a wire hoisting system and a motion capture system were built for experiments. The manipulator shape estimation experiment and obstacle avoidance trajectory planning experiment were conducted to verify the effectiveness of the algorithm on the physical object.