结构光辅助的惯性/视觉室内导航三维环境重构方法

3D environment restructure method with structured light for indoor vision/inertial navigation

室内环境全局信息对未知环境下的微型飞行器导航与定位具有重要的意义。针对惯性导航系统难以创建全局地图等问题,提出一种结构光辅助的惯性/视觉室内导航三维环境重构方法。该方法通过改进单点激光三角测距获得结构光上每个点在相机系下的坐标,使用惯性信息辅助图像匹配测角代替传统三维重建技术中的驱动单元获取两帧图像间角度信息,实现离散点组成的室内环境构造,结合平面构建策略剔除了错误离散点,改进了室内三维环境重构效果。实验结果表明:使用该方法获得的帧间角度准确,低转速下角度测量误差在5%以内;融合了离散点与平面的室内环境能够表现室内环境特征,并有效消除了错误离散点的干扰,将错误率降低到1%以下。

Global information of indoor environment is important for position and navigation of MAV（micro aerial vehicles） in unknown environment. To deal with the difficulty of building indoor environment of inertial navigation, a 3D environment reconstruction method with structured light is proposed. In the method, the laser triangulation ranging method is improved for calculating the coordinates of every structure＇s light points in camera coordinate. To replace the driver elements, the image matching assisted by inertial information is used to provide angular between pictures for construct the discrete points environment. A planes construction strategy is proposed to remove the error discrete points generated by interference. Combining with discrete points and planes, an accurate indoor 3D environment was reconstructed. Experiments show that： the angle got by image matching is accurate, the error of angular measurement is below 5%, the environment combined with discrete points and planes can express the characteristics of indoor environment accurately without wrong discrete points, and the error rate is reduced to less than 1%.