无人直升机自主着舰系统设计及仿真试验

设计了一套包含视觉目标识别单元、目标跟踪器、甲板状态跟踪器和模糊逻辑控制的无人直升机自主着舰系统.视觉系统通过对采集的图像预处理提取出可能的甲板标识区域,然后对此区域计算其仿射不变矩特征集合,对得到的特征集经过模糊识别技术实现对甲板降落区域的识别,并根据矩特征集计算出甲板运动状态信息.目标跟踪器采用了卡尔曼滤波技术,在目标二阶运动方程的基础上设计了跟踪算法.甲板运动状态跟踪器用于在无人直升机的降落期间实现对甲板尤其是触舰瞬间甲板的纵摇、横摇及浮沉状态的预测,该跟踪器采用了改进的卡尔曼多步预测算法.设计了基于模糊逻辑的比例积分控制器,由多个并行的对特定任务实施控制的子控制行为构成.仿真试验结果验证了所提出的算法.在仿真试验中,无人直升机能够正确的识别甲板目标区域,实现了对运动舰船在二维水平面内的跟踪,甲板状态跟踪器能够预测未来1~5个周期内甲板状态,控制器基于目标信息和甲板状态能够正确地引导无人直升机实现安全着舰.

Unmanned helicopter autonomous board landing system and simulation experiment

The unmanned helicopter autonomous landing system was composed of vision and recognition sytem, a target tracker, a deck-s state tracker and a fuzzy logic controller. The vision system performed to process the captured image and compute the image-s affine moment invariants. The label was recognized based on the fuzzy logic and those invariants. The label-s motion state was estimated based on the moment. A Kalman filter integrating into the control system was used to track the label and plan the route of unmanned helicopter. The board-s motion state, especially the pitch, roll and lift, is important for safe landing, so an improved Kalman forecast algorithm was designed to predict the board-s state of further one to five steps. The controller was customized based on behavior and fuzzy logic. Simulation results validate proposed algorithm. The unmanned helicopter can recognize the label accurately and track the label in any direction on ground. The improved Kalman filter can estimate the state future one to five periods, and there is less error than normal Kalman filters. The controller can generate the accurate commands and perform the autonomous landing on board safely.