作业型飞行机器人抓取后重心偏移的轨迹跟踪控制

Trajectory tracking control of center of gravity shift for aerial manipulator robot after grasping

中文关键词  作业型飞行机器人  重心偏移  滑模控制  自适应反演控制  小脑神经网络  

英文关键词  aerial manipulator robot  center of gravity shift  sliding mode controller  adaptive backstepping controller  the CMAC neural network

中文摘要

      作业型飞行机器人是指能够对环境施加主动影响的飞行机器人, 它通常由旋翼飞行器与机械臂组合而成. 本文针对作业型飞行机器人在动态飞行抓取后, 重心位置变化产生的系统控制难题, 设计了有效的跟踪控制策略. 首先, 在系统建模时引入重心偏移系统参数和重心偏移控制参数, 并考虑惯性张量不为常数, 提高了系统建模的精度. 然后, 在姿态解算时, 考虑重心偏移对系统性能的影响, 构建包含重心偏移系统参数的解算方法, 得到更高精度的期望翻滚角和期望俯仰角. 接着, 设计了基于滑模控制的重心偏移补偿位置控制器, 实现了有效的位置跟踪控制. 同时, 在姿态反演控制器的基础上, 加入自适应律估计重心偏移控制参数和变化的惯性张量, 再通过小脑神经网络逼近惯性张量的真实值, 提高姿态控制器的精度. 最后, 给出了所设计控制器的稳定性证明, 并在仿真环境下验证了所提出的方法的有效性和优越性.

英文摘要

      Aerial manipulator robot is a kind of flying robots that can exert active influences on the environments. It is usually a combination of a rotor craft and a robotic arm. Aiming at the system control problems of the center of gravity changing of the aerial manipulator robot after dynamic grasping, this paper design an effective tracking control strategy. At first, the center of gravity shift system parameters, the center of gravity shift control parameters and non-constant inertia tensor are introduced to improve the accuracy of system modeling. Then, considering to the effect of center of gravity shift on system performance, an attitude solution including the center of gravity shift system parameters is constructed to obtain a higher accuracy desired roll angle and desired pitch angle. Moreover, a sliding mode position controller is designed to compensate the center of gravity shift and and achieve effective tracking performance. Meanwhile, an adaptive law is introduced to estimate the center of gravity shift control parameters and non-constant inertia tensor based on a backstepping attitude controller. The Cerebellar Model Articulation Controller(CMAC) neural network is used to make the estimated inertia tensor closer to its true value. Finally, the stability analysis of proposed controllers are provided and the effectiveness and superiority of the proposed method are verified in a simulation environment. Meanwhile, an adaptive law is introduced to estimate the center of gravity shift control parameters and non-constant inertia tensor based on a backstepping attitude controller. The CMAC neural network is used to make the estimated inertia tensor closer to its true value. Finally, the stability analysis of proposed controllers are provided and the effectiveness and superiority of the proposed method are verified in a simulation environment.