【作者】 王旭；

【导师】 张军； 金伟明； 赵国普；

【作者基本信息】 东南大学 ， 电子信息（专业学位）， 2025， 硕士

【摘要】 包含柔性气囊结构的室内飞艇机器人具有能耗低、续航时间长、安全性高的优势,是一种特殊飞行器,在监测搜救、物流运输、人机交互等领域具有应用价值。当前飞艇机器人的研究主要集中在单个机器人的运动控制。本文受鸟群编队行为的空气动力学原理启发,开展飞艇机器人紧密跟随运动的建模与控制方法研究。主要研究内容为:(1)完成了飞艇机器人紧密跟随气动影响仿真与分析。研究了鸟群编队行为中的空气动力学机理,分别进行了双飞艇机器人紧密跟随飞行和单机飞行的空气动力学仿真。根据飞艇机器人的结构特点和不同飞行状态时的气流变化情况分析,分别建立了领航飞艇机器人和跟随飞艇机器人的气动影响模型,并进行了合理的模型简化设计。(2)完成了双飞艇机器人的运动学和动力学建模。建立了双飞艇机器人的坐标系、紧密跟随的运动学模型。进行了不同角色飞艇机器人的受力分析,并进行合理假设,结合领航飞艇机器人和跟随飞艇机器人的气动影响模型,分别得到了它们的动力学模型。(3)完成了飞艇机器人紧密跟随控制器设计与仿真。设计了双飞艇机器人紧密跟随运动控制方案。领航飞艇机器人采用串级PID控制算法,跟随飞艇机器人设计了PD-PID、线性混合PID、反步法和模型预测四种紧密跟随控制器。在MATLAB/Simulink中,针对跟随飞艇机器人的不同控制算法进行了直线、S曲线、圆形轨迹、螺旋线上升和Levy轨迹的紧密跟随运动仿真。根据仿真结果进行了不同控制器的控制性能分析和对比。(4)完成了飞艇机器人的样机设计和紧密跟随运动的实验研究。进行了基于动作捕捉定位系统的实验环境部署,设计了“X”型机架的飞艇机器人机械系统和硬件系统,基于STM32的飞艇机器人软件控制系统。分别进行了双飞艇机器人直线、S曲线、圆形轨迹、螺旋线上升和Levy轨迹紧密跟随飞行实验,实验结果表明设计的四种紧密跟随控制器都能实现三维空间中的紧密跟随运动。在常规轨迹中,线性混合PID控制的位置均方根误差在0.181m以内,模型预测控制的位置均方根误差在0.175m以内。在复杂Levy轨迹中,线性混合PID控制的位置均方根误差在0.225m以内,模型预测控制的位置均方根误差在0.194m以内。二者控制性能优于其他控制器,具有更稳定的紧密跟随控制效果。更多还原

【Abstract】 Indoor blimp robots with flexible airbag structures possess advantages of low energy consumption,long endurance,and high safety.As special aerial vehicles,they exhibit extensive application values in fields such as monitoring and reconnaissance,logistics transportation,and human-robot interaction.Current research on blimp robots has predominantly focused on the motion control of individual robots.Inspired by the aerodynamic principles of bird flocking behavior,this thesis conducts research on modeling and control methods for the close formation following flight of blimp robots.The specific research content is as follows:First,aerodynamic influence simulation and analysis for the close formation following flight of blimp robots are completed.The aerodynamic mechanisms in bird flocking behavior are studied,and aerodynamic simulations are performed for both dual-blimp close formation following flight and single-blimp flight.Based on the structural characteristics of blimp robots and airflow changes under different flight states,aerodynamic influence models for leader and follower blimp robots are established and reasonably simplified.Second,kinematic and dynamic modeling of dual-blimp robots is accomplished.Coordinate systems and kinematic models for close formation following flight are established.Force analyses of blimp robots in different roles are conducted with reasonable assumptions,and dynamic models for leader and follower blimp robots are derived by integrating interference models.Third,controller design and simulation for the close formation following flight of blimp robots are carried out.A control scheme for dual-blimp close formation following flight is designed,where the leader blimp robot employs a cascaded PID control algorithm.For the follower blimp robot,four types of close following controllers are designed:PD-PID,linear hybrid PID,backstepping,and model predictive.Simulations of close formation following flight along straight lines,S-curves,circular trajectories,spiral ascents,and Levy trajectories are conducted in MATLAB/Simulink for different control algorithms of the follower blimp robot.Control performance analyses and comparisons among different controllers are performed based on simulation results.Finally,prototype design and experimental research on the close formation following flight of blimp robots are completed.An experimental environment based on a motion capture positioning system is deployed.The mechanical and hardware systems of the blimp robot with an"X"frame and a software control system based on STM32 are designed.Close formation following flight experiments along straight lines,S-curves,circular trajectories,spiral ascents,and Levy trajectories are conducted for dual blimp robots.Experimental results show that the four designed close following controllers can all achieve close formation following flight in three-dimensional space.For conventional trajectories,the root-mean-square(RMS)position error of linear hybrid PID control is within 0.181 m,and that of MPC is within 0.175 m.For complex Levy trajectories,the RMS position error of linear hybrid PID control and MPC is within 0.225 m and 0.194 m,respectively.Both controllers outperform others in control performance,exhibiting more stable close formation following flight effects.更多还原