基于机器视觉技术的小型无人机飞行轨迹自动控制方法

Automatic control method for flight trajectory of small unmanned aerial vehicles based on machine vision technology

摘要:

小型无人机飞行过程中,面对复杂环境时的三维激光雷达采集模式容易出现出现“假回环”问题,由此生成的轨迹位置误差较大,导致小型无人机飞行轨迹自动控制质量下降,为此提出基于机器视觉技术的小型无人机飞行轨迹自动控制方法。通过无人机搭载相机获取无人机飞行区域环境图像,运用机器视觉技术完成色彩模型转化、曲率滤波、特征点提取与转化等图像处理工作,感知飞行环境中的障碍物分布信息,绘制出三维环境感知地图。机器视觉技术通过特征点提取和匹配算法,能够减少因激光雷达数据噪声或环境干扰导致的误匹配问题,这有助于降低“假回环”的出现概率。应用改进A*算法在机器视觉环境感知地图中搜索出一条飞行路线,为后续飞行轨迹控制提供重要的基础。结合迭代学习算法和PD控制算法建立自动控制方案,该控制方案主要考虑规划飞行路线和无人机实际飞行轨迹之间的误差,对飞行轨迹进行有效控制,保证无人机的飞行安全性。实验结果表明,依靠该方法完成飞行轨迹自动控制后,所呈现出的位置误差小于0.5m,能够更好地辅助小型无人机完成飞行任务,可以在实际中得到广泛应用。

Abstract:

During the flight of small unmanned aerial vehicles, the three-dimensional laser radar acquisition mode in complex environments is prone to the problem of "false loops", resulting in large errors in the generated trajectory position, leading to a decrease in the quality of automatic control of small unmanned aerial vehicle flight trajectories. Therefore, a machine vision based method for automatic control of small unmanned aerial vehicle flight trajectories is proposed. By using a camera mounted on a drone to capture images of the drone"s flight area environment, machine vision technology is applied to complete image processing tasks such as color model conversion, curvature filtering, feature point extraction, and transformation. This enables the perception of obstacle distribution information in the flight environment and the creation of a three-dimensional environment aware map. Machine vision technology, through feature point extraction and matching algorithms, can reduce the problem of false matching caused by LiDAR data noise or environmental interference, which helps to reduce the probability of "false loops". Applying the improved A * algorithm to search for a flight route in the machine vision environment perception map provides an important foundation for subsequent flight trajectory control. Establish an automatic control scheme by combining iterative learning algorithm and PD control algorithm, which mainly considers the error between the planned flight route and the actual flight trajectory of the drone, effectively controls the flight trajectory, and ensures the flight safety of the drone. The experimental results show that relying on this method to achieve automatic flight trajectory control results in a position error of less than 0.5m, which can better assist small unmanned aerial vehicles in completing flight tasks and can be widely applied in practice.