Abstract: Aiming at the real-time and precise positioning problem of UGV（unmanned ground vehicle） by UAV（unmanned aerial vehicle） in air-ground cooperative robots, a red double-circle positioning mark and mark recognition and positioning method are proposed. The combination of color segmentation and contour extraction is introduced in the method to reduce the number of extracted contour features and eliminate background information interference to reduce misrecognition. Then, a rapid detection algorithm for circular contours is proposed to quickly identify target contours and accurately locate the target pixel coordinates and direction. Finally, based on the pinhole camera imaging model, the three-dimensional coordinates and yaw angle of the target in the body coordinate system are estimated from the target pixel coordinates and direction. Experimental results show that when the relative height of the UAV and the UGV is 1.5 m, the positioning error of the method in the [x]-axis and [y]-axis directions is 3.9 mm and 3.6 mm, respectively, and the average processing time for each frame of image is 11.6 ms, which is better than 13.3 mm, 14.3 mm and 56.3 ms of the identification and positioning method based on kernalized correlation filter. Combining the method with UAV control, it can realize autonomous tracking and landing of UAV and improve the efficiency of air-ground cooperative robot, which has significant engineering significance.

Key words: air-ground collaborative robot, unmanned aerial vehicle（UAV）, real-time and precise positioning, positioning mark

摘要： 针对空地协同机器人中无人机对地面无人车的实时精准定位问题，提出一种红色双圆型定位标记及标记识别与定位方法。引入颜色分割与轮廓提取相结合的方式，减少提取到的轮廓特征数量，排除背景信息干扰以减少误识别；提出一种圆形轮廓快速检测算法，快速识别目标轮廓并准确定位目标像素坐标和方向；基于针孔相机成像模型，根据目标像素坐标和方向，估计出目标在机体坐标系下三维坐标和偏航角。实验结果表明，无人机与地面无人车相对高度1.5 m时，该方法在[x]轴和[y]轴方向定位误差分别为3.9 mm和3.6 mm，每帧图像平均处理耗时为11.6 ms，优于基于核相关滤波的识别定位方法的13.3 mm、14.3 mm和56.3 ms。该方法与无人机控制相结合，可以实现无人机协同跟踪与自主降落功能，提升空地协同机器人作业效率，具有显著的工程意义。

关键词: 空地协同机器人, 无人机, 实时精准定位, 定位标记