An adaptive proximal policy optimization algorithm based on mission-completion degree was proposed to address the challenge of maintaining large-separation formations for high-speed unmanned aerial vehicles (UAVs) in unknown and dynamic environments during manned-unmanned cooperative operations. An observation space that depended solely on relative state information was constructed, and a clipping-coefficient tuning mechanism driven by mission completion degree was devised. A segmented trajectory-truncation method was incorporated to accelerate convergence and enhance control stability. The simulation results indicate that relative to the baseline proximal policy optimization algorithm, marked improvements in convergence efficiency and tracking accuracy are achieved. Accurate formation is preserved even when the manned aircraft performs random maneuvers or abrupt heading changes, so the effectiveness and robustness of the approach in dynamic combat scenarios are validated, and an efficient solution for manned-unmanned formation keeping is provided.
所提出的基于任务完成度的自适应PPO算法(adaptive PPO based on task completion,APPO-TC)流程如下。其中,Vφ为当前价值网络参数φ对状态的价值估计;Vold为旧的价值网络对状态的价值估计;${\hat{A}}_{I}$为第I个样本的估计优势函数;αφ为价值网络参数φ的学习率;$ \nabla_{\phi}$为价值损失对价值网络参数φ的梯度。
针对PPO算法在实现无人机自主跟随有人机实现编队保持的问题,在PPO算法基础上进行了以下改进:通过建立基于相对距离和相对速度的状态空间,使得控制输入不依赖于全局信息,降低了对环境信息的要求,增强了系统的泛化性;通过任务执行完成度来调整裁剪系数ε的大小,以此来提高算法的收敛速度和稳定性;并在更新网络参数时引入轨迹截断设计,避免了轨迹太长带来的更新速度慢的影响。仿真结果表明,改进后的算法在3×103个训练回合内即可稳定收敛且收敛结果优于改进前的PPO算法;无人机能够对未知飞行轨迹完成编队保持任务,并且在有人机急转时慢慢保持到右侧 3 km 处的编队位置,证明该算法有一定的泛化性。在无人机自主跟随有人机实现编队保持问题上提出基于任务完成度的自适应PPO算法具有一定的创新性,对提高无人机编队保持的实现提供了新思路。在接下来的研究中,希望进一步改进算法使其具备规避航路障碍的功能并将算法拓展至复杂的三维空间。
WuJie, HuShenghua, QiaoShasha, et al. Research on combat concept and mission management technique of “Loyal Wingman” manned/unmanned aerial vehicle collaborative[J]. Avionics Technology, 2021, 52(2): 27-31.
LiuWenlin, PanZishuang, HanWei, et al. Research status and prospect of manned/unmanned aerial vehicle cooperative operation[J]. Journal of Naval Aeronautical University, 2022, 37(3): 231-241.
XiongWei, ZhangDong, RenZhi, et al. Research on intelligent decision-making methods forco ordinated attack by manned aerial vehicles and unmanned aerial vehicles[J]. Systems Engineering and Electronics, 2025, 47(4): 1285-1299.
HanWei, WuLiyao, ZhangYong. Review and prospect of flight control of carrier-based aircraft / unmanned aerial vehicles formation[J]. Science Technology and Engineering, 2019, 19(36): 73-80.
ZhangLu, ShaoZhengtu, WengChengxiang, et al. Research on application and key technologies of U.S. forces manned/unmanned aerial vehicle cooperative combat[J]. Tactical Missile Technology, 2022(6): 128-137.
LiLei, WangXianfeng, WangJi. Analysis on the latest development trend of manned-unmanned teaming of foreign forces[J]. Tactical Missile Technology, 2022(1): 113-119.
WuLiyao, HanWei, ZhangYong, et al. Formation keeping control for manned/unmanned aerial vehicle formation based on leader-follower strategy[J]. Control and Decision, 2021, 36(10): 2435-2441.
DuanKai, GaoHanwei, WangKe, et al. Cooperative control of UAV formation based on terminal sliding mode[J]. Journal of Naval Aviation University, 2023, 38(6): 483-489, 518.
LiuJinxin, ShengWei, ZhangYumin. Optimization algorithm for UAV formation consensus based on motion constraint[J]. Systems Engineering and Electronics, 2025, 47(1): 230-243.
WanRu, MaZiyuan, GongHuajun, et al. Design of unmanned aerial vehicle formation keeping controller based on improved consistency algorithmesign[J]. Journal of Beijing University of Aeronautics and Astronautics, 2025, 51(10): 3555-3566.
GouJinzhan, WuYu, DengJianing. Swarm intelligence and consensus theory based trajectory planning for acomplete flight of UAV formation[J]. Control and Decision, 2023, 38(5): 1464-1472.
ZhangBohan, ZhuXu, HuangJiangtao, et al. Consistent formation control of UAVs under wind interference and input constraints[J]. Journal of Naval Aviation University, 2024, 39(4): 745-754.
JiangTaimin, TanTai, LiHui, et al. Path following of 6-DOF fixed-wing UAV based on hierarchical deep reinforcement learning[J]. Computer Engineering, 2026, 52(4): 90-102.
YangZhipeng, LiBo, GanZhigang, et al. Path following of quadrotor UAV based on deep reinforcement learning[J]. Journal of Command and Control, 2022, 8(4): 477-482.
YangZhengquan, WangBingqian. Region formation control me-thod of UAVs with communication delay[J]. Science Technology and Engineering, 2025, 25(25): 10792-10799.
LiuXin, ZhangQianfei, LiuChengyu, et al. Moving target recognition and unmanned aerial vehicle following based on deep deterministic policy gradient[J]. Journal of Xi'an Polytechnic University, 2024, 38(4): 9-17.