Margareta Stefanovic

Distributed leaderless and leader-follower consensus of linear multiagent systems under persistent disturbances

The presence of unknown disturbances prevents agreement in multiagent systems. In this article, we propose observer-based leaderless and leader-follower consensus algorithms to achieve agreements in the presence of unknown unmatched persistent disturbances. Agents have linear timeinvariant dynamics and provide only relative output measurements. The simulation results verify the feasibility of the proposed ideas for these communication topologies.

Distributed optimal leaderless consensus of linear multiagent systems with polynomial state space models

Distributed consensus algorithms have been widely designed for linear time-invariant multiagent systems. However, multiagent systems are subject to unknown environments, which results in time-varying linear models. Additionally, it is desirable to investigate the consensus performance of the closed-loop MAS. In this article, we characterize the operating condition of the multiagent system by an unknown bounded independent parameter. We propose a state space model of the multiagent system where system matrices are polynomial functions of the independent parameter. Based on two different scenarios, we design two LQR-based optimal consensus algorithms that are robust with respect to the MASs varying operating condition. In each case, we calculate the final agreement value in the presence of modeling uncertainties and provide a sufficient condition in order to reach the state agreement on zero.

Distributed decoupling of linear multiagent systems with mixed matched and unmatched state-coupled nonlinear uncertainties

Distributed control algorithms have been widely designed to achieve consensus in a decoupled multiagent system. However, the presence of couplings may result in an unacceptable response of each agent because of the instability or poor transient performance of the overall closed-loop multiagent system. In this article, we consider a class of multiagent systems with linear nominal dynamics subject to the state-dependent nonlinear couplings. These nonlinearities appear as mixed matched and unmatched modeling uncertainties which are quadratically upper bounded. We investigate the case when the control-layer communication topology is different from the agent-layer coupling structure. We design optimal distributed control strategies that asymptotically decouple agents, and provide a guaranteed level of transient performance for the closed-loop coupled multiagent system. We further develop a fixed-gain linear quadratic regulator-based fully distributed decoupling system without global knowledge about the agent- and control-layer graph topologies.

Correction to: Results on the Robust Observer-Based Position Controller for Parallel Kinematic Machines

A correction is being made regarding Assumption 2: the assumption does not hold for the system under consideration, where only displacement measurements are available, but may hold under weaker conditions that allow partial velocities availability as well.

Distributed stabilization of linear multiagent systems with coupled state and input uncertainties

Using graph-theoretic tools, distributed control algorithms have been widely designed to manage a cooperative task among a group of individual agents. However, when agents are physically interconnected, these algorithms do not guarantee stability of the closed-loop multiagent system. In this paper, we consider a class of interconnected linear multiagent systems with state- and input-coupled modeling uncertainties. We assume that only few agents have access to their local state information, and others only provide lumped relative-state information. We propose a robust stabilization problem for an interconnected multiagent system with state- and input-coupled modeling uncertainties. By adding a virtual leader, possibly unstable, we reformulate this stabilization task to a leader-follower tracking problem. Incorporating the graph theory, we propose a linear quadratic regulator-based approach and propose a sufficient condition to systematically address this problem. Finally, we verify the feasibility of our approach through a simulation study.

A convex optimization method for improved coverage in Mobile Ad hoc Networks

Agents in Mobile Ad hoc Networks (MANET) typically have no knowledge of the environment that they are to be deployed in nor where they will be located. All information a mobile agent has access to is provided locally by other mobile agents within its communication range. Due to the nature of random deployment, agents may be clustered together which creates coverage holes in the network. Most of the existing work that focuses on improving area coverage deals with Wireless Sensor Networks (WSNs) where wireless connections between individual agents are not a concern when moving the agents. This work proposes a method to improve area coverage while preserving communication links between the mobile agents. A distributed consensus protocol based on Voronoi tessellation is introduced. A constrained convex optimization is then used to solve the distributed consensus protocol to determine an optimal position for a node inside of its Voronoi cell without losing communication with neighbors. This work is based on the belief that a node covers the most area when it covers most or all of its Voronoi cell. The convex optimization is then tested on a small network of agents and it is shown that area coverage is improved.