视觉无人机棚内煤场自主飞行与地图构建

Vision-based Unmanned Aerial Vehicle Autonomous Flight and Mapping in Indoor Coal Works

针对目前室内大型煤场煤储量估计方法中,固定位置激光打点方式存在盲区、不灵活、精度差,以及基于多旋翼无人机的方法难以适应无GPS的室内环境等缺点,提出结合视觉定位的无人机室内自主飞行盘煤方法。该方法通过融合5个方向的视觉信息,并结合无人机路径规划及避障算法,对煤堆进行了全覆盖的视觉成像,然后,通过运动推断结构方法进行三维建模,用于估计煤储量。经实验验证,所提方法有较好的室内定位精度,基于三维建模的煤堆储量估计与实际储量较为接近,证明了其有效性和可行性。

Aiming at the problems of the existence of fade area,inflexibility and low precision of fixed position laser measurement method and not adapting to GPS-denied environment of multi-rotors unmanned aerial vehicle（UAV）based method,a new method for indoor coal reserves estimation based on UAV autonomous flight with vision localization is proposed during the research of coal reserves estimation in large indoor coal works.Full coverage imaging of coal stockpile is completed by combining path planning and obstacle avoidance with visual odometry which fuses visual information in five directions.Then the threedimensional reconstruction of coal stockpile could be obtained by using structure from motion（SfM）method for coal reserves estimation.The experimental results show that our method can achieve good localization precision and the estimated coal reserves based on mapping is close to the actual value,which verifies feasibility and effectiveness of our method.