With the swift advancement of unmanned aerial vehicle (UAV) technology, quadcopters have shown promising applications in military reconnaissance, environmental monitoring, and other fields. However, existing UAV precision position control methods suffer from collision risks, potential loss of control, and insufficient control accuracy. This study introduces a precise position control method for quadcopters that integrates improved active disturbance rejection control with multi-sensor fusion. active disturbance rejection control is improved by introducing a linear extended state observer with time-varying gains. A dynamic weighted Kalman filter strategy is employed to combine multi-source sensor data to enhance the accuracy and disturbance rejection performance of the quadcopter’s precise position control. The experimental results show that in terms of position tracking, the proposed model significantly outperforms the proportional-integral-derivative (PID) controller and the traditional active disturbance rejection control in terms of the root mean square error (RMSE) of the three axes; in the scenario of GPS signal loss, the proposed model significantly reduces the positioning error compared to the traditional fusion method, and the convergence speed of disturbance estimation is significantly improved. This study effectively improves the control accuracy and disturbance rejection performance of quadcopters in complex environments, providing important technical support for the conception and utilization of highly reliable UAV systems.
This study addresses the coupled optimization of task allocation, task visiting order, global route planning, and local obstacle avoidance for low-altitude multi-UAV missions. A low-altitude hierarchical cooperative hybrid intelligent optimization algorithm, termed LA-HCHIOA, is developed by coupling GA-based population evolution, ACO-based subsequence reconstruction, VNS-based local refinement, and hierarchical path-cost feedback. The collaborative tasks of UAVs are dynamically allocated through task-UAV feasible matching, task sequence coding and path cost feedback. Path planning achieves obstacle avoidance and multi-aircraft conflict coordination through global route generation and local rolling correction, and incorporates task priority, path length, energy consumption, safety interval and load balancing into a unified optimization process. The experimental results showed that the total path length of the proposed method was 15,784.3 m, the fleet energy consumption was 798.7 kJ, the task completion time was 584.9 s, the task completion rate was 99.2
This paper presents a dynamic modeling and adaptive control framework for a cyclocopter propelled by cycloidal rotors. The simulation model integrates the kinematics of the cycloidal rotor mechanism, a reduced-order blade-element aerodynamic formulation with an unsteady correction, and the six-degree-of-freedom rigid-body dynamics. To address the nonlinear and periodically varying aerodynamic characteristics, a deep deterministic policy gradient (DDPG)-based adaptive PID controller is proposed, in which the trained Actor performs state-dependent online scheduling of 18 PID gains for the position and attitude control loops. Simulation studies evaluate aerodynamic characteristics, thrust-vectoring yaw control, trajectory tracking under wind disturbances and parametric uncertainties, and generalization to unseen trajectories. The proposed DDPG-PID controller achieves performance comparable to the baseline controllers under nominal conditions, while providing lower overall RMSE, IAE, and ITAE under disturbed conditions. The trained policy also successfully tracks unseen circular and lemniscate trajectories without retraining. Online gain histories demonstrate state-dependent PID gain scheduling, while five independent training runs show reasonably consistent tracking performance. Within the adopted simulation model, these results indicate that the proposed DDPG-PID framework can improve the adaptability and disturbance-rejection capability of cyclocopter trajectory control.
This paper presents a distributed model predictive control (DMPC) scheme for quadrotor swarms that keep a V-shape formation in cluttered, dynamic environments. The vehicles form a hierarchical leader–follower structure and exchange predicted trajectories over the planning horizon. Each local problem, solved by sequential quadratic programming (DMPC-SQP), balances formation keeping, sibling consensus, avoidance of static and moving obstacles, and control effort; rotor downwash enters through an anisotropic inter-vehicle safety region. Formation deformation and restoration follow from the trade-off between the safety and formation costs, with no switching logic. With a terminal cost built from an LQR law, a warm start with a solver safeguard and a consistency constraint on the broadcast trajectories, the analysis establishes recursive feasibility and, under a horizon condition, practical asymptotic stability of the formation after the last obstacle passage, without assuming optimal solves; collision avoidance is promoted by soft penalties and is not guaranteed. In a seven-vehicle scenario with a narrow gap, a tube corridor, a crossing moving obstacle and two mid-flight joins, no collision occurred in the simulated runs and the V reformed to within 1 mm, while a reactive potential-field baseline collided and used 7.9 times more control effort.
The time-varying aerodynamic coefficients and the inadequacy of accurate path tracking during the range flight are remarkable problems in the flight management of aircraft. To deal with the path planning of a fixed-wing aircraft, this paper researches the input configuration on the dynamic equation of aircraft by cost function optimization and feedback control, to fulfill path tracking at the preset initial states. Firstly, the dynamic equation of aircraft is set up by Newton’s law, on both translation and revolving channels. As the aerodynamics are time varying, the multi-variable interpolation with regards to aerial angles and control surface deflection is proposed to formulate aerodynamic coefficients at the arbitrary instant. Then variable search over the configured objective function is employed to find optimal input in view of the initial conditions on the path segments; besides, the effect of damping mode in basis functions is proven crucial to reduce state chattering and increase system stability. Furthermore, the closed-loop feedback control is applied on partial channels, which together with input optimization strengthens the tracking accuracy of system on the preset path conditions. Finally, numerical simulation on the aircraft system is undertaken, which validates that the proposed input-search algorithms are capable of strengthening the effect on path tracking in time-varying flight compared to PID control.
