| In the last two decades,cooperative control of multi-agent systems has received extensive attention due to its wide range of applications in multi-UAV formation flying,multi-robot cooperative exploration,distributed optimization,and multi-sensor coverage.Leader-follower cooperative tracking control,as one of the foundations of the above research,aims to design distributed control methods that enable all followers to track the leader reference trajectory.However,multi-agent systems inevitably face the influence of multiple uncertainties,mainly including measurement uncertainties due to sensor systematic errors and stochastic errors,model uncertainties due to insufficient modeling accuracy,and input uncertainties due to non-ideal actuator dynamics.The above multiple uncertainties seriously affect the cooperative tracking control performance of multi-agent systems.Based on the above analysis,this dissertation focuses on the robust cooperative tracking control of leader-follower multi-agent systems under multiple uncertainties,and carries out a systematic research on the impact of the above uncertainties by constructing a hierarchical cooperative control architecture based on distributed observers and decentralized tracking controllers.The main innovations and contributions of this dissertation are as follows:1.A robust cooperative tracking control method based on the compensation of sensor measurement error is proposed address the problem that the followers are affected by the measurement uncertainties caused by the sensor errors.Firstly,based on a non-smooth distributed observer,an estimation of the input and states of the leader is achieved by fitting the input of the leader’s augmented system using a non-smooth function,and then the reference tracking trajectory information for each follower is formed.Secondly,in the decentralised tracking controller,the follower measurement signals are pre-processed by a low-pass filtering in order to inhibit the high-frequency stochastic errors,and then for the phase lag introduced by low-pass filtering and the residual low-frequency systematic errors,the phase lag compensator and the measurement error estimator are proposed respectively,in which the online estimation and active compensation of the phase lag and residual errors are realized by using the closed-loop model information for solving the phase lag and residual errors to improve the robustness of the control system and the tracking performance of the leader reference trajectory.Based on the Lyapunov stability theory,the bounded convergence of the closed-loop system tracking errors under the proposed cooperative control method is demonstrated,and the numerical simulation results verify the effectiveness and performance advantages of the method.2.A robust cooperative tracking control method based on barrier function-based disturbance observer(BFDO)is proposed to address the problems of unknown dynamic model of the leader and the followers suffering from model uncertainties.Firstly,an adaptive distributed observer based on neural network is utilized to fit the unknown dynamic function of the leader,which achieves an effective estimation of the leader’s states.Secondly,a decentralised robust tracking controller combining the BFDO with feedback control is proposed based on the back-stepping method,which uses the known model information and the barrier function to fit the follower’s model uncertainties,and achieves an active estimation and compensation of these uncertainties to improve the robustness of the control system and the tracking performance of the leader reference trajectory.Based on the Lyapunov stability theory,the bounded convergence of the closed-loop system tracking errors under the proposed cooperative control method is proved,and the numerical simulation results verify the effectiveness and performance advantages of the method.3.A robust cooperative tracking control method based on uncertainty and disturbance estimator(UDE)and dynamic surface control(DSC)is proposed for the problems of unknown input of the second-order leader and the followers being affected by model uncertainties under non-ideal actuator dynamics.Firstly,by analyzing the stability conditions of the classical UDE-based control system under non-ideal actuator dynamics,it is revealed that the actuator dynamics characteristics limit the range of the UDE parameters,which affects the estimation performance of the UDE against uncertainties and the robustness of the control system.Secondly,based on a second-order fixed-time distributed observer,the convergence of the observation error is accelerated by introducing a nonlinear power term to achieve a fixed-time estimation of the leader states.Finally,based on the back-stepping method,a decentralised robust tracking controller combining UDE and DSC is proposed,in which the limitation of the actuator dynamic characteristics on the UDE parameters is eliminated by using the actuator model information,the structural constraint that the UDE only applies to matched uncertainties is satisfied by decomposing the higher-order follower system with mismatched uncertainties into the lower-order subsystems with only matched uncertainty through back-stepping,and the “complexity explosion”problem is avoided by using first-order low-pass filtering to replace the inherent differentiation operation of the back-stepping method through DSC.Based on the small gain theorem and the Lyapunov-like theorem,the bounded convergence of the closedloop system tracking errors under the proposed cooperative control method is proved,and the simulation results of the full attitude cooperative tracking control of quadrotor UAVs verify the effectiveness and performance advantages of the method.4.A robust cooperative tracking control method based on a high-order sliding mode disturbance observer(HOSMDO)is proposed to address the problems of unknown input of the third-order leader and the follower being subjected to model uncertainties under nonideal actuator dynamics.Firstly,based on a third-order fixed-time distributed observer,the observation error reaches the sliding mode surface in a finite time and thereafter converges to zero in a fixed time through the sliding mode control,which achieves a fast estimation of the leader states.Secondly,based on the holistic design method,a decentralised robust tracking controller combining the HOSMDO and the feedback control is proposed,in which the holistic feedback control design avoids the cumbersome recursion process of the back-stepping method,and finite-time estimation as well as active compensation of the follower uncertainties and their derivatives are carried out by using of the known model information and the higher-order sliding mode differentiation technique to improve the robustness of the control system and the tracking performance of the leader reference trajectory.Based on the homogeneous theory and Lyapunov stability theory,the asymptotic convergence of the closed-loop system under the proposed cooperative control method is demonstrated,and the simulation results of the full-attitude cooperative tracking control of quadrotor UAVs verify the effectiveness and performance advantages of the method. |
