27 July 2022 in IEEE/ASME Transactions on Mechatronics
In this article, we propose a novel finite-memory-structured online training algorithm (FiMos-TA) for neural networks to identify and predict the unknown functions and states of an unmanned aerial vehicle (UAV). The proposed FiMos-TA is designed based on a system reconstructed by accumulating the states from the UAV dynamics. The system is redefined by replacing the unknown nonlinear functions of the UAV with neural networks, and a random walk modeling is adopted to design a training algorithm. The proposed FiMos-TA with a finite memory structure updates the weights of the neural network by accumulating the refined measurements of a UAV on the receding horizon. The training law of the proposed FiMos-TA is obtained by introducing the Frobenius norm and confirms a robust performance against modeling uncertainties and identification errors. The robustness and accuracy of the proposed FiMos-TA are verified through experiments.
23 March 2022 in IEEE Transactions on Network Science and Engineering
This paper investigates the adaptive asymptotic tracking control for networked nonlinear stochastic systems. Different from having the necessity of prior knowledge of the unknown control coefficients in the conventional adaptive control of nonlinear stochastic systems, in this study, the limitation of control coefficients in the stability analysis is relaxed by constructing a new Lyapunov function that contains the lower bounds of the control gain function. By constructing a smooth function with a positive time-varying integral function and utilizing the boundary estimation method, asymptotic tracking control can be guaranteed. At the same time, for nonlinear stochastic systems with unknown control coefficients, a neural adaptive event-triggered strategy that greatly saves communication resources while ensuring system performance is proposed. Finally, simulation results show that the proposed control scheme can guarantee the realization of the control objectives.
28 June 2024 in IEEE Transactions on Industrial Electronics
In this article, a novel distributed finite memory online learning (DFMOL) strategy is proposed for accurately identifying the dynamic models of multiple unmanned aerial vehicles (UAVs). The proposed DFMOL derives a recurrent neural network (RNN) model for each UAV, based on which it redesigns a distributed measurement model that reflects the information and connectivity of neighboring UAVs. The proposed DFMOL strategy stacks the distributed measurements on a receding horizon and estimates the weights of each RNN using only a finite number of distributed measurements, ensuring robust performance based on the characteristics of finite memory. The gain of learning law is derived by establishing the Frobenius norm minimization problem to minimize the impact of disturbances, systematic uncertainties, and identification errors from UAVs and their neighbors. Furthermore, the proposed DFMOL strategy is derived iteratively to reduce the computation for real-time hardware implementation. Real-time experiments are carried out to show the robust and accurate performance of the proposed distributed identification.
7 January 2022 in IEEE Access
Mobile robot localizations have been extensively studied, and various algorithms for multiple-robot localization have been developed. However, existing methods for multiple-robot localization often exhibit poor performance under harsh conditions, such as missing measurements and sudden appearance of obstacles. To overcome this problem, this paper proposes a novel method for multiple-robot localization in wireless sensor networks. The proposed method is theoretically based on the finite memory estimation and utilizes relative distance and angle measurements between robots. Thus, the proposed method is referred to as distributed finite memory estimation from relative measurements (DFMERM). Due to the finite memory structure, the DFMERM has inherent robustness against computational and modeling errors. Moreover, the novel distributed localization method using relative measurements shows the robustness against missing measurements. Robust DFMERM localization performance is experimentally demonstrated using multiple mobile robots under the harsh conditions.
26 July 2024 in IEEE Transactions on Industrial Electronics
This article proposes a finite-memory-based front-wheel angle estimation (FMFWAE) strategy for robust and accurate performance in a steer-by-wire (SBW) system. The SBW system integrates the steering actuator and vehicle models, considering both the steering angle and vehicle motion. Our proposed FMFWAE, designed based on the SBW system, operates with the finite-memory property, using only a limited number of recent measurements and control inputs. This finite-memory property provides robustness to the proposed FMFWAE against linearization errors, model uncertainties, and fault signals. Furthermore, by developing a Frobenius norm minimization problem, we obtained the gain matrices of FMFWAE to minimize the impact of linearization errors and model uncertainties on the SBW system. The robust and accurate performance of FMFWAE was validated through real-time experiments.
2 August 2023 in IEEE Signal Processing Letters
The finite memory estimation based learning algorithm (FME-LA 1) is a newly developed technique for identifying unknown nonlinear systems that demonstrates remarkable accuracy and robustness against disturbances and error accumulation by utilizing a finite number of measurements on a moving window. However, there are several drawbacks to FME-LA 1. First, the weights of higher-order or complex systems cannot be learned because it only learns the output layer's weights. Second, it does not consider noise statistics, resulting in significant performance degradation in complex environments. Finally, it is not suitable for real-time implementation because the computation increases rapidly with the size of the moving window. To address these limitations, we propose the moving window iterative nonlinear system identification (MWI-NSID) scheme, which guarantees robustness and accuracy with a significant reduction in computation. In simulation results on a complex nonlinear system, the proposed MWI-NSID scheme demonstrated robustness, accuracy, and significantly reduced computation.