Guangming Xie

Controlling anonymous mobile agents with unidirectional locomotion to form formations on a circle

Abstract We study the circle forming problem in which a group of anonymous mobile agents are required to form a formation when moving on a given circle. The agents are constrained to move in the one-dimensional space of the circle only in the counterclockwise direction, but not the opposite way. Distributed, cooperative, sampled-data control strategies are designed that only take nonnegative values. We prove that the multi-agent system under such constrained control input can be guided to reach the prescribed circle formation asymptotically with the additional guarantee that no collision between agents ever takes place. The theoretical analysis is further validated through simulations.

Brief paper: Periodic stabilizability of switched linear control systems

Stabilizability via direct/observer-based state feedback control for discrete-time switched linear control systems (SLCSs) is investigated in this paper. For an SLCS, the control factors are not only the control input but also the switching signal, and they need to be designed in order to stabilize the system. As a result, stabilization design for SLCSs is more complicated than that for non-switched ones. Differently from the existing approaches, a periodic switching signal and piecewise constant linear state feedback control are adopted to achieve periodic stabilizability for such systems. It is highlighted that multiple feedback controllers need to be designed for one subsystem. For discrete-time SLCSs, it is proved that reachability implies periodic stabilizability via state feedback. A necessary and sufficient criterion for periodic stabilizability is also established. Two stabilization design algorithms are presented for real application. Moreover, it is proved that reachability and observability imply periodic stabilizability via observer-based feedback for discrete-time SLCSs. Periodic detectability, as the dual concept of periodic stabilizability, is discussed and the stabilization design algorithms via observer-based feedback are presented as well.

An Adaptive and Online Underwater Image Processing Algorithm Implemented on Miniature Biomimetic Robotic Fish

Abstract This paper proposes a novel underwater image processing approach for different targets detection and preliminary positioning of a miniature robotic fish with low computational resources. A key aspect of our approach is that the image processing algorithm is performed online, i.e., while swimming, rather than as an offline process. Another key aspect is that the algorithm distinguishes the targets adaptively within large distance range. Specifically, considering the poor quality of underwater image as well as the low computational resources of the robot, a real-time image preprocessing algorithm, named accelerated automatic color equalization (AACE) model, is first adopted to improve color contrasts and sharpness of borders of the image. Further, HSV-based adaptive threshold segmentation, lump detection, reflection elimination and landmark matching are executed successively in a robust way for locating the target and calculating the distance between the robot and target. A series of relevant experiments with the robotic fish are conducted systematically to demonstrate that the proposed image processing approach is adaptive, efficient and real-time for online applications of the vision-based miniature swimming robots.

Emergence of leadership in a robotic fish group under diverging individual personality traits

Variations of individual’s personality traits have been identified before as one of the possible mechanisms for the emergence of leadership in an interactive collective, which may lead to benefits for the group as a whole. Complementing the large number of existing literatures on using simulation models to study leadership, we use biomimetic robotic fish to gain insight into how the fish’s behaviours evolve under the influence of the physical hydrodynamics. In particular, we focus in this paper on understanding how robotic fish’s personality traits affect the emergence of an effective leading fish in repeated robotic foraging tasks when the robotic fish’s strategies, to push or not to push the obstacle in its foraging path, are updated over time following an evolutionary game set-up. We further show that the robotic fish’s personality traits diverge when the group carries out difficult foraging tasks in our experiments, and self-organization takes place to help the group to adapt to the level of difficulties of the tasks without inter-individual communication.

Antiphase Formation Swimming for Autonomous Robotic Fish

Abstract This paper proposes distributed control laws for formations of swimming robotic fish generating antiphase sinusoidal body waves. The control laws are inspired by the mathematical model for the hydrodynamics of schools of cruising fish, which reveals that fish swimming in diamond-shape formations with synchronized antiphase body waves can benefit greatly from energy saving. The phase dynamics of the body waves of the robotic fish are modeled by coupled Kuramoto oscillators and the stability analysis for the phase dynamics are carried out for coupling topologies corresponding to diamond-shape formations. It is proven that the body waves can be synchronized in specific antiphase patterns. Simulations and experiments further validate the effectiveness of the proposed control laws.

Analysis and modeling of multi-rate networked control systems with long time delay

By multi-rate networked control systems, we mean the sampling periods of the sensor, the controller and the actuator in networked control systems are not the same, that is to say there are more than one sampling rates in networked control systems. The models of multi-rate networked control systems with no network-induced delay and long time delay are proposed respectively in this paper. It is stated that the multi-rate networked control systems with long time delay can be formulated as a discrete-time switched stochastic system. Especially the multi-rate networked control systems with long time delay can be formulated as a discrete-time switched deterministic system when the actuator is clock driven and the state noise is not considered. An example is given in the end.

Modified snowdrift games for multi-robot water polo matches

In this paper, using game theoretic ideas, we propose a modified snowdrift game to study the emergence and evolution of cooperation among robots in multi-robot water polo matches. We first formulate a game setting in which groups of robots play matches repeatedly that are modeled by modified snowdrift games. Then by introducing a cooperation coefficient we investigate how evolutionary stability is affected by cooperation efficiency through replicator dynamics in infinite populations. We identify evolutionarily stable strategies (ESS) under different fixed values of the cooperation coefficient. We further study the co-evolution of the cooperation efficiency with game dynamics using simulations of Fermi processes in finite populations. It is found that the cooperation efficiency improves when robots are capable to learn.

Speed estimation for robotic fish based on pressure sensor

Based on pressure sensor, a speed estimation method for robotic fish is presented in this paper. Based on property of pressure sensor, the relation model between pressure and speed is fitted by experimental analysis. Using this relation model, the swimming speed of robotic fish can be calculated by pressure. A validation experiment was conducted with a freely swimming robot. The experiment results demonstrated the feasibility and accuracy of this measure method, which providing a new way to measure the swimming speed of robotic fish.

Observer-based containment control of multi-agent system with linear dynamics

This paper considers the containment control of linear multi-agent system with weakly connected information topology. By defining the node-induced strongly connected subgraph as strong component, the agents contained in strong components with zero in-degree are called leaders and the others are called followers. Then, a sufficient and necessary condition is proposed for all agents to achieve containment by employing an observer-based protocol, where all followers finally converge to the convex hull formed by leaders and each strong component of leaders asymptotically achieve consensus. Since the number of leader agents in each strong component are always larger than one, the proposed scheme is more robust for the malfunction of leaders. Further, two LMIs are given to design the observer-based protocol. Numerical examples are included to illustrate the validity of the proposed results.