基于视觉伺服的欠驱动无人水面艇自主靠泊方法

Autonomous Berthing Method of Underactuated Unmanned Surface Vehicle Based on Visual Servo

无人水面艇在开阔水域的自主航行得到了广泛应用,但由于码头结构复杂、船只较多等因素,无人水面艇的靠泊主要还需人工来完成,这已成为无人水面艇全自主作业的瓶颈。文章提出一种基于视觉伺服的无人水面艇自主靠泊方法。首先,通过无人水面艇搭载的视觉系统采集泊位的场景,从中提取出泊位标志物并将其作为视觉跟踪的对象;然后,利用标志物图像和期望图像的几何参数计算出航向偏差角,通过标志物图像计算标志物和无人艇的位置关系并求得虚拟航线,进而得到偏航距离;最后,选取航向偏差角和偏航距离作为控制变量,控制器实时调整无人水面艇航向和航速使其驶向泊位,当标志物图像与期望图像像素面积的差值小于设定阈值时即停止自主靠泊任务。实船试验结果表明,在较小风浪条件下靠泊位置误差小于0.7 m,航向误差小于12°。

The autonomous navigation has been widely used to unmanned surface vehicles(USV) in open waters. Due to the complex structure of docks and the large number of vessels near the berthing area, the berthing of USV still needs to be completed manually by operators, which has become the bottleneck of full autonomous operation of USV. The paper proposes a visual servo-based autonomous berthing method for USV. Firstly, the scene of berth is collected through the visual system carried by the USV, from which the berth markers are extracted and used as the object of visual tracking. The heading deviation angle is calculated by using the geometric parameters of the marker image and the objective image. The heading deviation distance from USV to the virtual route is obtained geometrically with the marker image. Finally, both the heading deviation angle and distance are chosen as the controlled variables for control design, which aims at driving the USV to the expected berthing destination by adjusting the heading and speed of the USV in real time. The autonomous berthing task is completed when the difference between the marker image and the expected image is less than a threshold value. The result of USV test shows that the berthing position error of less than 0.7 m and the heading error of less than 12° are achieved under the given experimental condition when the wind and waves are small.