Cooperative localization and control of networked systems

Networked systems have attracted recurring research interests from the control community due to its extensive applications in civil and military areas. Most research interests on networked systems revolve around two fundamental problems, that is, localization and control, where localization serves as the underpinning of control strategies. Despite the technological advances on localization of networked systems, there is still in need of a general localization framework that can accommodate different kinds of measurements for localization, especially for distributed localization. In addition, there are two challenging problems in control of networked systems. The first is to steer a group of agents to maneuver as a whole so that they can dynamically respond to environment changes. The second is to steer a group of agents to capture a dynamic object. Motivated by these issues, for localization of networked systems, this thesis studies both centralized and distributed localization with multi-sensor fusion. For control of networked systems, this thesis focuses on distributed formation maneuver control and cooperative pursuit. To localize a mobile robot (free node) in both 2-D and 3-D spaces, a general centralized localization framework based on graph optimization approach is proposed, which can accommodate different kinds of measurements with varying measurement time intervals. Special emphasis will be on range-based localization. Range and trajectory smoothness constraints are constructed in a position graph, then the robot trajectory over a sliding window is estimated by a graph-based optimization algorithm. Moreover, convergence analysis of the algorithm is provided, and the effects of the number of iterations and window size in the optimization on the localization accuracy are analyzed. Extensive experiments on micro unmanned aerial vehicles (UAV) under a variety of scenarios verify the effectiveness of the proposed algorithm and demonstrate a much higher localization accuracy than the existing range-based localization methods, especially in the altitude direction. Although the proposed centralized method can achieve accurate localization for a free node in a small-scale network, it may not be applicable to localize multiple free nodes in a large-scale network due to communication bandwidth and power constraints. To solve this problem, a general distributed localization framework is proposed, where the local coordinate frames of different nodes have different unknown orientations. An angle-displacement rigidity theory is developed to explore algebraic condition and graph condition of network localizability. Then, a distributed localization protocol is proposed, which can globally estimate the locations of all the free nodes of a network if the network is infinitesimally angle-displacement rigid. The proposed method unifies local-relative-measurement-based distributed localization approaches. A remarkable advantage is that the proposed method can be applied in both generic and non-generic configurations with an unknown global coordinate frame in both 2-D and 3-D spaces. Besides the localization problems in networked systems, we also study control problems in networked systems. To steer a group of agents to maneuver as a whole so that they can dynamically respond to environment changes, distributed formation maneuver control algorithms are designed for multi-agent systems, which can steer a team of mobile agents to achieve desired collective maneuvers so that geometric pattern, translation, orientation, and scale of formation can be changed continuously. Unlike most existing results requiring generic or convex nominal configurations, the proposed control algorithms can be applied to either non-generic or non-convex nominal configurations. It is worth noting that the closed-loop tracking errors converge to zero globally. In addition, to steer a group of agents (pursuers) to capture a dynamic object (evader), a distributed pursuit algorithm is proposed to enable pursuers to form an encirclement and approach the evader, where the evader has faster speed than that of the pursuers and is allowed to move freely without any constraint. Considering that forming an encirclement and approaching the evader are two independent processes, a trade-off algorithm is proposed to balance between forming an encirclement and approaching the faster evader. Moreover, sufficient capture conditions are derived based on the initial spatial distribution and the speed ratios of the pursuers and the evader. Simulation and experimental results on ground robots verify and validate the effectiveness and practicability of the proposed method.Doctor of Philosoph