AUV与水下运动平台进行动态对接的导引控制技术研究

During the process of mobile docking between an Autonomous underwater vehicle and a moving submarine, there are many related difficulties. It is difficult to estimate the motion state of the target because of weak positioning observation. It is also difficult to construct multi-constraint model and calculate a optimal reference speed. Besides, the non-uniform and time-varying current field brings difficulties to the sensing and control system. Based on the time sequenced positioning results of USBL single beacon, a method aims to solve the problem of absolute motion state estimation within a relatively wide range. To improve rapidity and accuracy of under water target feature recognition and visual positioning, a underwater linear light source array with unique feature is put forward and a corresponding artificial intelligence based rapidly recognizing method is developed to realize real time and high precision positioning at short range. To ensure the acoustic and vision positioning accuracy, the collision safety and the docking timeliness, space-time constraint model problem was discussed and a guidance method that calculates optimal reference speed vector dynamically was came up with. To observe and compensate the disturbance of a non-uniform and time-varying flow field induced by the moving submarine, we presents a non-model-dependent fast identification method and the steepest control method. Finally, the adopted navigation, guidance and control methods are verified by the semi-physical simulation of the AUV and the moving submarine. Through theoretical research and experimental study, this project will provide good methods for the safely, reliably and efficiently docking between an underactuated AUV and a moving submarine in the future.

智能水下机器人（AUV）与运动的平台进行动态对接面临弱观测条件下目标运动状态估计难、多约束条件下的最优参考速度规划难、对非均匀时变流场的观测和补偿控制难的问题。本项目拟基于时序声学单信标定位数据实现较大范围时的绝对运动状态估计；以水下光学特征快速识别为出发点提出具有唯一性特征的水下线型光源阵列结构以及人工智能快速识别与测量技术实现近距离的实时、高精度定位；针对声/光定位精度、碰撞安全性以及对接时效性等时空多约束实现最优参考速度矢量在线规划（制导）；针对AUV受主艇运动产生的非均匀、时变流场扰动影响，基于人工智能方法提出非模型依赖的海流干扰力快速感知识别和最速补偿控制方法；最后通过半物理仿真试验验证所提出的导引、规划和控制技术。通过理论研究和试验研究，为未来欠驱动AUV与水下运动平台的安全、可靠、高效对接提供导引控制方法。

AUV自主对接回收技术可实现AUV随时自动充电并在其对接成功后采取有线方式高速传输数据,可大大提高AUV任务执行效率。另外，与固定式对接回收技术相比，移动式对接回收技术具有更大的灵活性且AUV依靠移动式对接回收技术能与其他无人航行器合作完成更复杂的任务。例如，当AUV执行远距离海洋考察任务时，水面无人艇可以作为AUV的运输平台节省AUV能源，并且能辅助AUV定位使其采集到的数据更加准确。由于AUV移动式对接技术具有重大的研究意义和实用价值，所以本项目首先介绍了欠驱动AUV移动式对接任务流程，研究了考虑运动模型和终端约束的路径规划算法，仿真结果证明欠驱动AUV移动式对接归航路径规划算法在任意初始相对位姿条件下欠驱动AUV都能计算出满足运动学约束和终端约束条件归航路径；然后研究了结合降阶线性扩展状态观测器和反步法的轨迹跟踪控制方法，仿真结果证明了观测器可有效地估计和补偿未知扰动，且在时变海流影响下欠驱动AUV仍能较好地跟踪期望轨迹；最后提出了基于回转圆的轨迹规划方法，用回转实验数据代替欠驱动AUV动力学方程，并认为移动式对接规划轨迹由直线和回转圆组成，AUV在规划轨迹上采用最大速度航行，半物理仿真实验证明了该方法的可行性和实时性；另外还进行了陆上双目视觉学定位实验，试验结果显示在测距小于20m的范围之内双目视觉定位的坐标精度与测距之比均在3%以内，而双目视觉定位的姿态角精度均在15°的范围以内，尤其是在距离小于5m时，该双目视觉定位的坐标精度与测距之比小于1%，姿态角的误差范围在6°以内。这说明该双目视觉定位算法的定位精度较高，满足AUV末端导引对接的要求。