The 2017 Asian Control Conference (ASCC 2017) Workshop, Gold Coast, Australia
Advances in distributed control and formation control systems
Background and motivation
In recent years cooperative coordination control and distributed control systems have gained considerable attention in the control community. This has been motivated by various applications such as formation control, unmanned air vehicles, coordination in complex networks, sensor networks, distributed optimization, etc. The central themes in the study of cooperative control for networked multi-agent systems include the understanding of the role of cooperation, the mechanism of information sharing between distributed agents, the stability and achievement of a global task arising from local interactions, coordination and cooperation with measurement constraints, and the robustness against measurement/communication perturbations, among others.
Formation control, which is one of the most actively studied topics within the realm of distributed cooperative control, generally aims to drive multiple agents to achieve prescribed constraints on their states. Roughly speaking, formation control aims to design distributed controllers such that a group of spatially distributed agents could reach some pre-defined formations involving geometric relationships between them. Such geometric relationships can be described by relative positions, bearings, distances, or a mix of different geometric variables, depending on the context and control requirements.
In the recent decade there has been much advancement in formation control systems, including different techniques in formation control, bearing rigidity theory and its application in formation systems and network localization, stability and convergence of formation systems, robustness properties with practical considerations. However, there are still several unsolved problems, such as a complete understanding of formation systems involving nonlinear control laws, or the fundamental limitations or trade-offs between local controllers and a global formation task.
This workshop aims to bring together active researchers in this field to showcase their latest achievements of techniques, designs and applications, as well as new directions of formation control and distributed cooperative control arising in various engineering systems.
Date and venue
December 17, 2017 (a full day workshop), at Gold Coast, Australia
Please see here for the workshop agenda.
Zhiyong Sun, Australian National University; Minh Hoang Trinh, Gwangju Institute of Science and Technology (GIST)
We have identified the following speakers who will contribute to this workshop:
Speaker: Prof. Brian D. O. Anderson
Affiliation: Research School of Engineering, Australian National University, Australia; Hangzhou Dianzi University, Hangzhou, China
Title: Open problems in rigidity with application to formation control
Abstract: This talk will cover two topics: a possible route to establishing a Laman type theorem for three-dimensional joint-bar frameworks, and a statement of a problem with conjecture on rigidity for a collection of separate rigid bodies in an ambient three-dimensional space, with directional constraints known between them. The problem has arisen in an application context of GPS-denied cooperative localisation of agents in a formation, using direction only sensing.
Speaker: Prof. Hyo-Sung Ahn
Affiliation: School of Mechanical Engineering, Gwangju Institute of Science and Technology, South Korea
Title: Gradient-based formation control: Complete solutions and open problems
There are mainly two different classes in distributed formation control problems: Distance-based approaches and bearing-based approaches. In this talk, we would like to present existing solutions and some unsolved issues in distance-based approaches under gradient-based approaches. Due to highly nonlinear characteristics of gradient-based approaches, we focus on simple K3 and K4 graphs. However, although the problems for K3 have been completely solved, the problems for K4 are considered quite challenging. It is even tougher to solve the problems of K4 - one edge.
Speaker: A./Prof. Shaoshuai Mou
Affiliation: College of Engineering, Purdue University, USA
Title: A Distributed Algorithm for Least Square Solutions of Linear Equations
Abstract: This talk will present a discrete-time distributed algorithm for achieving least square solutions in multi-agent networks, in which each agent only knows part of the overall linear equation and receives information from its nearby neighbors. The algorithm does not involves any small step-size, converges exponentially fast for fixed-undirected networks.
Speaker: Prof. Daniel Zelazo
Affiliation: Faculty of Aerospace Engineering, Israel Institute of Technology
Title: Formations over Directed Graphs and Local Coordinate Frames
Abstract: The implementation of coordination and control algorithms for multi-robot systems often depends explicitly on the sensing modalities available to the robots. Common coordination goals, such as formation control, can have vastly different control algorithms depending on what information is available to each robot. One of the current challenges in formation control deals with robots operating harsh environments, and in particular where knowledge of a common reference frame is unknown. This problem is further made difficult by sensors (such as cameras) that are mounted on the frame of the robot and thus are only able to sense other robots from their local frames, leading to a directed sensing network. In this talk, we explore on approach to solving this problem based on an extension of rigidity theory for configurations of robots in SE(2). We discuss how formation shapes can be determined using bearing measurements in local coordinate frames and propose a scale-free formation control law under these sensing constraints. The talk will conclude with a discussion on remaining open challenges in this arena.
Speaker: Minh Hoang Trinh
Affiliation: School of Mechanical Engineering, Gwangju Institute of Science and Technology, South Korea
Title: Bearing-Based Formation Control of A Group of Agents with Leader-First Follower Structure
The topic of this talk is bearing-based formation control of a group of autonomous agents with the leader-first follower (LFF) structure in an arbitrary dimensional space. Firstly, the bearing-based Henneberg construction and some properties of the LFF formation are introduced. Then, we propose and analyze bearing-only control laws which almost globally stabilize LFF formations to a target formation. Further strategies to rotate and to rescale the target formation are also discussed. Finally, simulation results are provided to support the analysis.
Speaker: Dr. Zhiyong Sun
Affiliation: Research School of Engineering, Australian National University, Australia
Title: Conservation laws and invariance principles in networked coordination control systems
Abstract: In this talk we discuss several conservation laws in formation control systems. Specifically, we reveal conservations of linear momentum and angular momentum for gradient-based multi-agent formation systems modelled by single integrators, and show several corresponding conservation/decay laws for double-integrator formation stabilization systems and double-integrator flocking systems, respectively. By exploiting translation and rotation symmetry properties and insights from Noether's theorem, we further establish a multi-agent version of the relation between symmetry and conservation laws for gradient-based coordination systems derived from general potential functions, from which we generalize the conservation/decay laws to more general networked control systems.
We further study coupled dynamical systems and establish several invariance principles relating to the dimensions of the subspace spanned by solutions of each individual system. We consider two types of coupled systems, one with scalar couplings and the other with matrix couplings. Via the rank-preserving flow theory, we prove that scalar-coupled dynamical systems possess the dimensional-invariance principles, in that the dimension of the subspace spanned by the individual systems’ solutions remains invariant. For coupled dynamical systems with matrix coefficients or couplings, necessary and sufficient conditions are given to characterize dimensional-invariance principles.
Speaker: Dr. Hector Garcia de Marina
Affiliation: Ecole Nationale de l'Aviation Civile (ENAC), Toulouse, France
Title: Shape and motion control of formations under inconsistent measurements
The gradient-based control is a very popular tool in the literature for stabilizing formations of autonomous agents in a distributed way. Although the exchange of information among neighbouring agents is not often required by this kind of controllers, the 'undirected nature' of the sensing graph requires a pair of neighbouring agents to have exactly the same measurement about the distance that separates them. Often it is an unavoidable situation to have two sensors, such as laser scanners, that give different outputs for the same physical distance due to noises or biases. These inconsistent measurements induce undesired collective motions and a distortion in the desired shape of the formation.
The first part of this talk will show how to systematically design and implement different local estimators that removes such inconsistent measurements among neighbouring agents governed by single and double integrators. A discussion about the benefits and disadvantages of each local estimator will be provided and insights of this technique in actual robotics systems will be given as well.
The second part of the talk is devoted to the reverse-engineering of the inconsistent measurements such that the formation can achieve not only the desired shape, but a desired stationary motion, e.g., translation, rotation and/or scaling of the shape. In particular, we will provide an open-source software tool written in Python that given a formation and a desired motion as inputs, it provides the set of motion parameters to be included in the employed gradient-based controller.
Speaker: Prof. Zhiyun Lin
Affiliation: Hangzhou Dianzi University, Hangzhou, China
Title: A graph Laplacian approach to distributed localization and formation control
Abstract: Distributed multi-agent coordination has been a very active subject studied extensively by the systems and control community in last decades, including distributed consensus, formation control, network localization, distributed optimization, etc. This talk is going to present our recent work on distributed localization and formation control based on a unified graph Laplacian approach. By introducing Laplacian matrices with complex entries and signed-value entries, distributed coordinate-free strategies are proposed to solve the localization problems with different types of inter-agent measurements (range, bearing, or relative position) and the formation control problem with different types of formation specifications. It has been shown that the developed approach ensures global convergence of the algorithms.