In this study, an online adaptive optimal control scheme is developed for solving the infinite-horizon optimal control problem of uncertain non-linear continuous-time systems with the control policy having saturation constraints. A novel identifier-critic architecture is presented to approximate the Hamilton–Jacobi–Bellman equation using two neural networks (NNs): an identifier NN is used to estimate the uncertain system dynamics and a critic NN is utilised to derive the optimal control instead of typical action–critic dual networks employed in reinforcement learning. Based on the developed architecture, the identifier NN and the critic NN are tuned simultaneously. Meanwhile, unlike initial stabilising control indispensable in policy iteration, there is no special requirement imposed on the initial control. Moreover, by using Lyapunov's direct method, the weights of the identifier NN and the critic NN are guaranteed to be uniformly ultimately bounded, while keeping the closed-loop system stable. Finally, an example is provided to demonstrate the effectiveness of the present approach.

This study is concerned with stochastic stability of a new extended filtering for non-linear systems subject to measurement packet losses. The measurements sensored are transmitted to the estimator through a packet-dropping network. By introducing a time-stamped packet arrival indicator sequence, the measurement loss process is modelled as an independent, identically distributed (i.i.d.) and therefore a Bernoulli process. The boundedness of estimation error covariance matrices is proved by showing the existence of a critical threshold for measurement packet arrival probability. It is also shown that, under appropriate assumptions, the estimation error remains bounded as long as the noise covariance matrices and the initial estimation error can be ensured small enough. Finally, simulation results validating the effectiveness of this proposed filtering framework are also presented.

This study presents a novel reduced-order non-linear observer for vehicle velocities estimation based on vehicle dynamics and Unified Exponential tire model. Yaw rate is chosen to construct the reduced-order observer since it can be conceived as the function of vehicle velocities. The observer is designed such that the error dynamics system is input-to-state stability (ISS), where model errors including mass and CoG variation, and estimation or measurement error of the maximum tire–road friction coefficient are considered as additive disturbance inputs. Then, the condition of the observer gain satisfied is obtained by the ISS analysis and the lower observer gain is obtained through the convex optimisation described by the linear matrix inequalities. The proposed observer requires fewer tuning parameters and thus indicates an easier implementation compared with the existing extended Kalman filter. Simulation results demonstrate the effectiveness of the proposed reduced-order non-linear observer, which is also validated through experimental data from Hongqi vehicle HQ430. Furthermore, its computational efficiency is shown based on the laboratory Field Programmable Gate Array and System on a Programmable Chip testing platform.

A systematic identification method for non-linear stochastic spatiotemproal (SST) systems described by non-linear stochastic partial differential equations (SPDEs) is investigated in this study based on pointwise observation data. A theoretical framework for a semi-finite element model approximating to an infinite-dimensional system is established, and several fundamental issues are discussed including the approximation error between the underlying infinite-dimensional dynamics and the model to be identified, and its rationality etc. Based on the proposed theoretical framework, a general identification method with irregular observation data is provided. These results not only provide an effective method for the identification of non-linear SST systems using measurement data (both offline and online), but also demonstrate a potential solution for the analysis, design and control of non-linear SST systems from a numerical point of view.

This study presents an asynchronous algorithm for distributed constrained optimisation problems in networks of agents. The iterative optimisation algorithm maintains a local estimate at each node and depends on local gradient or gradient-like updates in combination with a consensus policy, where an agent averages its own value with a current or outdated value of another. This asynchronous scheme does not require that agents exchange state information frequently, so it is more energy-efficient and more realistic than the synchronous one. Moreover, the proposed algorithm is fully distributed, that is, all agents only share data with their neighbours through local broadcasts. The proposed algorithm is applied to a distributed regression problem with robustness to outliers in sensor networks. Simulation results are provided to demonstrate the validity and superiority of the proposed scheme.

In this study, a novel indirect iterative learning control (ILC) strategy is presented for a robotic manipulator that performs repeat operation and is also subjected to non-Gaussian disturbances. The performance index about the entropy of tracking error and the related optimisation method are used to update the local parameters of controller between any two adjacent batches. Moreover, the entropy is employed as it is a unified probabilistic measure of uncertainty quantification regardless whether the random disturbances are Gaussian distribution or not. Thus, an innovative performance index about tracking error entropy that represents the relationship between the entropy of error and controller gains is proposed in order to obtain the controller which can drive the uncertainty of output error as small as possible with the increase of the batch number. Then the non-linear stochastic optimal method is presented so as to obtain updated gains for the next batch. Stability of closed-loop system is analysed. Finally, a comparison between a classic ILC and the proposed approach is given. Moreover, the effectiveness and feasibility of the proposed control schemes is verified by some simulation results of robotic trajectory tracking.

This study considers a consensus problem of discrete-time high-order multi-agent systems with digraph. A concept r-consensus for discrete-time multi-agent systems is introduced, and consensus protocols are proposed for solving such problems. A sufficient and necessary condition for r-consensus is obtained using matrix theory. As a special case, criteria of r-consensus for third-order systems are given, in which the exact relationship of feedback parameters is established. Finally, the effectiveness of the theoretical results is demonstrated through a numerical simulation.

This study investigates the synchronisation problem of identical linear high-order multi-agent systems. A new dynamical controller is constructed by the internal model approach, which does not depend on the controller state information of neighbouring agents, but only on the weighted sum of relative output errors and the local measured output. The proposed controller can work for some classes of agents having unstable modes and always work well for the agents having all its eigenvalues in the closed left-half complex plane. A simulation example illustrates the efficacy of the analytic results.