为解决在全球定位系统(Global Positioning System，GPS)信号缺失和环境感知不足的条件下，只使用单目相机进行可重构浮体厘米级精准对接的实际问题，本文提出了一种用于受到外部干扰的可重构浮体拼接过程的鲁棒非线性模型预测控制(Nonlinear Model Predivtive Control，NMPC)策略。本文首先定义相机平面中的3 个图像矩作为视觉特征获得图像运动学模型，然后通过整合该图像运动学模型和浮体的动力学模型，并建立了基于图像的视觉伺服(Image Based Visual Servo，IBVS)系统模型。根据此模型为浮体的拼接设计了一种NMPC控制方案，从而保证了浮体各物理约束和摄像机的视界约束。为了改善系统鲁棒性，根据系统连续性假设设计了一种收紧状态约束来处理外部干扰。最后，通过仿真实验和湖试实验验证了该方案的有效性。
Nonlinear Model Predictive Control for Reconfigurable Marine Body Docking Based on Visual Servoing
This paper proposed a novel disturbance rejection robust nonlinear model predictive control(NMPC)framework for the docking of reconfigurable marine bodies based on visual feedback control,to solve the practical problem of using a monocular camera for centimeter level precise docking under the conditions of Global Positioning System(GPS)signal loss and insufficient environmental perception.The proposed controller relies on an image-based visual servoing system model which integrates two components of the image kinematics and reconfigurable marine body dynamics.The former one is derived to capture the kinematics of reconfigurable marine body.This is made possible by using three image moments as visual features in the virtual camera plane to describe the movement of real camera mounted on the marine body.To guarantee the satisfaction of physical constraints of marine body as well as visibility constraint of visual targets,the MPC scheme is developed based on image-based visual servoing model.The tightened state constraints are further applied to compensate the external disturbances,thereby improving the robustness of the proposed MPC.Extensive simulation and lake test shows that the effectiveness and efficiency of the control scheme.
万方数据
维普
