Contents I. Introduction
Dynamic target tracking tasks are commonly encountered in fields such as peacekeeping and counter-terrorism. In many instances, such as in mountainous or urban complex environments, unmanned aerial vehicles (UAVs) are required to generate feasible trajectories in real-time based on the state of dynamic targets and the surrounding environment. However, obstacles in the environment and the high dynamics of the target may lead to sensor failures in tracking the target in real-time, resulting in target loss and an inability to carry out subsequent tasks. Therefore, the real-time planning of appropriate tracking trajectories based on the state of the environment is of great practical importance in complex settings. There has been significant research on UAVs or other mobile robots performing dynamic gaming or tracking tasks [1], [2]. Most of this work, however, focuses on target tracking in obstacle-free environments [3], [4]. When executing dynamic target tracking tasks in complex environments, obstacles pose increased challenges to quadrotor flight safety, dynamical feasibility, and trajectory smoothness. Xi L [5] proposed a guidance time-optimal model predictive control framework to generate quadrotor pursuit trajectories, using finite-jump methods to find time-optimal finite-jump trajectories. Lai S [6] presented a quadrotor online trajectory planning framework with boundary state constraints, combining neural networks and model predictive control (MPC) to assess the trajectories to meet the demands of real-time application in cluttered settings. Xi L and Peng Z [7] developed a quadrotor trajectory generation method based on optimal smooth B-splines for tracking moving targets in complex environments, considering relative tracking modes or limited fields of view of onboard sensors. Compared to most methods, their solution take into account the safety flight zones, physical limits of the airframe, and smoothness, ensuring flight safety, dynamical feasibility, and tracking performance. Inspired by the aforementioned studies, this paper investigates the issue of UAV dynamic target tracking in complex environments and proposes a trajectory planning method based on differential evolution for UAVs to track dynamic targets under complex conditions. This method concurrently considers factors such as UAV flight stability, safety, and line-of-sight maintenance, generating a finite-jump trajectory curve to guide the UAV in effectively tracking dynamic targets.
