Coordinated Control For Multi-agent Systems With Constraints

With the rapid development in communication technology, network theory and computer science, researchers in the field of system and control science have extended the investigation from a single dynamical system to multiple coupled systems. Coordinated control for multi-agent systems is such a case in point, which means that the whole system reaches a coordinated motion only relying on local interactions among agent neighbors. The advantage has effectively promoted its application in transportation, economy, military, engineering and other fields. In practice, multi-agent systems always suffer many negative impacts from the real environment, while time-delay, including communication delay and input delay, and limited resources are concerned in this paper. Algorithms are respectively proposed to solve the problems of distributed estimation and control for mobile sensor networks with coupling delays and sampled-data based event-trigger strategy for consensus of multi-agent systems with input-delay.Mobile sensor network, as an individual multi-agent system, has more advantages in environment adaptability than static wireless sensor network due to its mobility feature. But the benefit comes with increased risk in collision. Moreover, communication delay is a significant factor in terms of information exchange among agents. This paper presents an approach combining Kalman-Consensus filtering and flocking control, which can not only overcome the influence of communication delay, but maintain connectivity in mobile sensor network. By tools of Lyapunov theorem, matrix theory and algebraic graph theory, a sufficient condition on the system stability is derived and feedback coefficients are constrained by a delay-dependent inequality.Event-trigger strategy implies that communication unit and actuator unit in multi-agent systems are activated only when a well-designed event occurs, which obviously reduces the resource cost. Usually, the event is generated when a threshold is broken. With the consideration of extending the limitations in existing literatures and dealing with the input-delay and nonlinear dynamics, this paper studies the sampled-data based event-trigger strategy for consensus of multi-agent systems. Each agent is allocated with a task of storing specific information, and the trigger condition is designed in a much different way because newly defined measurement errors of neighbors' and leader' s are contained. The contribution is that not only the continuous communication between neighbors is avoided but each agent's controller is triggered only at its own event time instants. Base on Lyapunov-Krasovskii method and LMI tool, system stability is proved, and feedback matrix in control input and trigger matrix in trigger condition are solved.