基于DSST的四旋翼目标视觉跟踪系统研究

Research on visual tracking system of quadrotor target based on DSST

摘要:

随着图像识别技术的快速发展，实现复杂动态环境下无人机的高精度目标跟踪已成为智能飞行控制领域的关键课题。针对高空强扰动场景中的小目标跟踪难题，将判别尺度空间跟踪（DSST）算法与四旋翼无人机运动控制系统深度融合，通过构建高度自适应PID 控制器实现动态速度调节，结合速度指令均值滤波算法，有效解决了传统方法在三维空间跟踪中的响应迟滞与运动抖动问题。结果表明：优化后的控制系统可在20 m飞行高度下实现三自由度（俯仰、偏航、升降）协同跟踪控制，对较小特征尺寸的移动目标保持5 m/s 的稳定追踪速度；本文提出的控制策略为无人机在应急救援、边境巡查等复杂场景中的目标持续追踪提供了可靠的技术方案。

Abstract:

With the rapid development of image recognition technology, achieving high-precision target tracking for unmanned aerial vehicles (UAVs) in complex dynamic environments has become a key issue in the field of intelligent flight control. This paper addresses the challenge of small target tracking in high-altitude strong disturbance scenarios by deeply integrating the Discriminative Scale Space Tracker (DSST) algorithm with the motion control system of a quadrotor UAV. By constructing a highly adaptive PID controller to achieve dynamic speed regulation and combining it with a speed command mean filtering algorithm, the response lag and motion jitter problems in traditional methods for three-dimensional space tracking are effectively solved. Experiments show that the optimized control system can achieve three-degree-of-freedom (pitch, yaw, and altitude) coordinated tracking control at a flight altitude of 20 meters and maintain a stable tracking speed of 5 m/s for moving targets with small feature sizes. The control strategy proposed in this study provides a reliable technical solution for the continuous tracking of targets by UAVs in complex scenarios such as emergency rescue and border patrol.