Category: IEEE Transactions on Automatic Control

Ensuring safety is crucial for the successful deployment of autonomous systems, such as self-driving vehicles, unmanned aerial vehicles, and robots acting close to humans. While there exist many controllers that optimize certain criteria, such as energy consumption, comfort, or low wear, they are usually not able to guarantee safety at all times for constrained nonlinear systems affected by disturbances. Many controllers providing safety guarantees, however, have no optimal performance. The idea of this article is, therefore, to synthesize a formally correct controller that serves as a safety net for an unverified, optimal controller. This way, most of the time, the optimal controller is in charge and leads to a desired, optimal control performance. The safety controller constantly monitors the actions of the optimal controller and takes over if the system would become unsafe. The safety controller utilizes a novel concept of backward reachable set computation, where we avoid the need of computing underapproximations of reachable sets. We have further developed a new approach that analytically describes reachable sets, making it possible to efficiently maximize the size of the backward reachable set. We demonstrate our approach by a numerical example from autonomous driving.

We consider a setting where two noncooperative players optimally influence the evolution of an initial spatial probability in a game-theoretic hierarchical fashion (Stackelberg differential game), so that at a specific final time the distribution of the state matches a given final target measure. We provide a sufficient condition for the existence and uniqueness of an optimal transport map and prove that it can be characterized as the gradient of some convex function. An important by-product of our formulation is that it provides a means to study a class of Stackelberg differential games where the initial and final states of the underlying system are uncertain, but drawn randomly from some probability measures.

This article develops a distributed solution to the fully heterogeneous containment control problem (CCP), for which not only the followers’ dynamics but also the leaders’ dynamics are nonidentical. A novel formulation of the fully heterogeneous CCP is first presented in which each follower constructs its virtual exosystem. To build these virtual exosystems by followers, a novel distributed algorithm is developed to calculate the so-called normalized level of influences (NLIs) of all leaders on each follower, and a novel adaptive distributed observer is designed to estimate the dynamics and states of all leaders that have an influence on each follower. Then, a distributed control protocol is proposed based on the cooperative output regulation framework, utilizing this virtual exosystem. Based on the estimations of leaders’ dynamics and states and NLIs of leaders on each follower, the solutions of the so-called linear regulator equations are calculated in a distributed manner, and consequently, a distributed control protocol is designed for solving the output containment problem. Finally, theoretical results are verified by performing numerical simulations.

The consensus control of multiagent systems (MASs) over randomly switching communication topologies has drawn increasing attention from researchers, since the mean square consensus or the most surely consensus is significant and practical. This manuscript focuses on the output consensus problem for heterogeneous MASs with input saturation constraints over the Markovian randomly switching topologies. The main challenge of solving the concerned problem lies in the interplay among the heterogeneous dynamics, input saturation constraints and the Markovian randomly switching topologies. To overcome such a challenge, a class of distributed adaptive observers is first designed for all agents to deal with the uncertainty caused by the randomly switching topologies. Then, a class of local state observers is presented for estimating each agent’s state information. Based on the above steps, a class of antiwindup controllers is constructed and the control gain matrices are selected using only the dynamics information of each node. Finally, the effectiveness of the consensus protocol is demonstrated.

We propose decentralized primal-dual methods for cooperative multiagent consensus optimization problems over both static and time-varying communication networks, where only local communications are allowed. The objective is to minimize the sum of agent-specific convex functions over conic constraint sets defined by agent-specific nonlinear functions; hence, the optimal consensus decision should lie in the intersection of these private sets. Under the strong convexity assumption, we provide convergence rates for suboptimality, infeasibility, and consensus violation in terms of the number of communications required; examine the effect of underlying network topology on the convergence rates.

This article is concerned with minimizing the average of $n$ cost functions over a network, in which agents may communicate and exchange information with each other. We consider the setting where only noisy gradient information is available. To solve the problem, we study the distributed stochastic gradient descent (DSGD) method and perform a nonasymptotic convergence analysis. For strongly convex and smooth objective functions, in expectation, DSGD asymptotically achieves the optimal network-independent convergence rate compared to centralized stochastic gradient descent. Our main contribution is to characterize the transient time needed for DSGD to approach the asymptotic convergence rate. Moreover, we construct a “hard” optimization problem that proves the sharpness of the obtained result. Numerical experiments demonstrate the tightness of the theoretical results.

During the implementation of a cooperative algorithm, information about the agents’ velocity may be unavailable due to the space constraint and availability of sensors. Thus, it gives rise to the design of distributed average tracking (DAT) algorithms without using agents’ velocity measurements. These are denoted as velocity-free DAT problems. The existing literature has addressed such problems in the presence of an undirected graph for the reference signals with bounded position, velocity, and acceleration differences. We propose a velocity-free DAT algorithm under a weight-unbalanced strongly-connected digraph that represents the most general network structure for achieving DAT. Additionally, the proposed algorithm works for a broader range of time-varying references, having bounded acceleration differences among themselves. Linear stability theory is used to establish uniform ultimate boundedness of the errors for bounded acceleration differences. Asymptotic convergence of the errors is guaranteed for converging acceleration differences. Unlike the existing works, our DAT algorithm does not need any update law for the gains. Thus, the approach is computationally efficient. Numerical simulations with the comparison with the state-of-the-art demonstrate the performance of our algorithm over a wider range of time-varying references under weight-unbalanced graph.

Approximating model predictive control (MPC) policy using expert-based supervised learning techniques requires labeled training datasets sampled from the MPC policy. This is typically obtained by sampling the feasible state space and evaluating the control law by solving the numerical optimization problem offline for each sample. Although the resulting approximate policy can be cheaply evaluated online, generating large training samples to learn the MPC policy can be time-consuming and prohibitively expensive. This is one of the fundamental bottlenecks that limit the design and implementation of MPC policy approximation. This technical article aims to address this challenge, and proposes a novel sensitivity-based data augmentation scheme for direct policy approximation. The proposed approach is based on exploiting the parametric sensitivities to cheaply generate additional training samples in the neighborhood of the existing samples.

Advertisement.

Prospective authors are requested to submit new, unpublished manuscripts for inclusion in the upcoming event described in this call for papers.