动态贝叶斯网络下多无人机目标协同搜索控制研究

Research on Multi UAV Target Collaborative Search under Dynamic Bayesian Network

摘要:

在复杂环境连续域内因特征稀疏容易丢失无人机目标定位,出现多无人机目标协同搜索路径重合现象,导致冲突风险系数增加,为此对基于动态贝叶斯网络的多无人机目标协同搜索控制方法进行了研究。在无人机上安装高分辨率图像传感器设备获取目标图像,通过Lucy-Richardson算法对于目标图像进行运动模糊补偿后,利用动态贝叶斯网络提取目标形状、移动特征,采用特征匹配方法进行多无人机目标定位。根据目标位置与环境的图像信息构建栅格地图,在栅格地图中通过计算搜索优先级分配搜索目标,将分配目标作为搜索路线中的关键路径点,从而进行多无人机目标协同搜索控制路径规划,通过控制器不断调整飞行轨迹以实现对多无人机目标协同搜索控制。实验测试结果表明,设计方法的目标漏搜率比较低,碰撞冲突风险系数低,多无人机目标协同搜索控制成功率较高,由此证明该方法在多无人机目标协同搜索控制方面具有明显优势。

Abstract:

In complex environments with sparse features in continuous domains, it is easy to lose drone target localization, resulting in the phenomenon of multiple drone target collaborative search paths overlapping, leading to an increase in conflict risk coefficient. Therefore, a multi drone target collaborative search control method based on dynamic Bayesian networks has been studied. Install high-resolution image sensor devices on drones to obtain target images, perform motion blur compensation on the target images using the Lucy Richardson algorithm, extract target shape and movement features using dynamic Bayesian networks, and use feature matching methods for multi drone target localization. Construct a grid map based on the image information of the target location and environment, allocate search targets in the grid map by calculating search priority, and use the allocated targets as key path points in the search route to carry out multi drone target collaborative search control path planning. The controller continuously adjusts the flight trajectory to achieve collaborative search control of multiple drone targets. The experimental test results show that the target omission rate of the design method is relatively low, the collision conflict risk coefficient is low, and the success rate of multi drone target collaborative search control is high. This proves that the method has significant advantages in multi drone target collaborative search control.