具有视觉导航和目标识别的小型多功能无人驾驶直升机

【作者】 聂振钢； 张晓林； 李宏伟； 陆国雷；

【Author】 zhengang.nie,zxl,lihongwei.luguolei

【机构】 北京航空航天大学电子信息工程学院；

【摘要】 本文提出了一种基于任务的导航策略,采用了图像传感器、DGPS、惯性传感器、超声波传感器等多传感器融合技术,实现对无人直升机的状态估计及实时的导航控制。采用基于形态学和图像测量的图像定位和识别方法,意在抓住图标内部最为本质的几何特性和拓扑结构信息,同时避免了模板匹配中图像归一化等繁琐操作,对于尺度、旋转、光线变化不敏感,获得很高的识别正确率和定位精度。更多还原

【Abstract】 Our UAV team proposed an task oriented navigation policy. The difference between our system and traditional navigation system is that we take advantage of the multi-sensor fusion technology to combine the data from image sensor, DGPS, INS and ultrasonic sensor and to implement the status estimation and navigation control of helicopter.The morphologic method for image location and recognition we introduced, intend to abstract the innate geometric and shape information of each label, meanwhile it avoided some complicated operation of template matching , such as normalization, etc. The algorithm is not sensitive to scaling, rotation and lighting changes, and has a high rate of correct recognition and poison precision.更多还原