旋翼飞行机械臂系统的混合视觉伺服控制

Hybrid visual servoing for rotor aerial manipulation system

旋翼飞行机械臂是一种具有强耦合特性的机器人系统,借助视觉进行自主作业还存在诸多问题,如实时深度估计、目标极易丢失以及目标笛卡尔空间模型重建等.本文针对传统的基于图像与基于位置的视觉伺服的缺陷以及系统自身欠驱动等问题,建立了运动学模型和提出了基于力平衡原理的动力学联合建模,并通过欧几里得单应性矩阵分解设计出旋翼飞行机械臂系统的混合视觉伺服控制方法,在图像空间控制平移、笛卡尔空间控制旋转,减弱了平移与旋转之间的相互影响实现解耦效果,改善了系统对非结构因素的抗扰性能和全局稳定性.通过仿真和实验检验了系统鲁棒性和算法优越性.

Unmanned aerial manipulations are special robotic systems which exists strong coupling characteristics.There are many issues using the vision for actively manipulate in environment,such as real-time depth estimation,the target is extremely easy to lose and the target Cartesian space model reconstruction.W e formulate the kinematic and joint dynamic equations of the system based on force balance principle,which could deal with the respective drawbacks of classical image-based and position-based visual servoing as well as the matter of the system itself under-driven.Then,Through the decomposition of Euclidean homography matrix,proposing a hybrid visual servoing control scheme for unmanned aerial manipulation,which controls pan in image space and controls rotation in Cartesian space to achieve decoupling effect,is proposed to improve the systems interference immunity to non-structural factors and global stability.Finally,simulation and experimental results validate the robustness of system and demonstrate the superiority of algorithm.