George M. Mathews

Decentralised decision making in heterogeneous teams using anonymous optimisation

This paper considers the scenario where multiple autonomous agents must cooperate in making decisions to minimise a continuous and differentiable team cost function. A distributed and asynchronous optimisation algorithm is presented which allows each agent to incrementally refine their decisions while intermittently communicating with the rest of the team. A convergence analysis provides quantitative requirements on the frequency agents must communicate that is prescribed by the structure of the decision problem. In general the solution method will require every agent to communicate to and have a model of every other agent in the team. To overcome this, a specific subset of systems, called Partially Separable, is defined. These systems only require each agent to have a combined summary of the rest of the team and allows each agent to communicate locally over an acyclic communication network, greatly increasing the scalability of the system.

Decentralised Optimal Control for Reconnaissance

This paper introduces a decentralised negotiation algorithm, where multiple distributed decision makers can propose and refine future decisions to optimise a common team objective function. Based on theoretical convergence results, a practical algorithm is developed. The information requirements of the system, that is, what is each platform required to know, are examined and a scalable cooperative control algorithm proposed. This is applied to the control of a reconnaissance or information gathering task consisting of multiple UAVs.

Scalable Decentralised Control for Multi-Platform Reconnaissance and Information Gathering Tasks

This paper describes the general multi-platform reconnaissance or information gathering task. It is shown that the general problem must be solved in a centralised manner using information supplied by every platform. However, if for a specific problem the objective function is partially separable it is shown the optimal control problem can be solved in a decentralised fashion for an arbitrary sized group. This is demonstrated for scenarios with Gaussian probabilities and an indoor mapping problem

Scalable Decentralised Decision Making and Optimisation in Heterogeneous Teams

This paper considers the scenario where multiple autonomous agents must cooperate in making decisions to minimise a common team cost function. A distributed optimisation algorithm is presented. This allows each agent to incrementally refine their decisions while intermittently receiving updates from the team. A convergence analysis provides quantitative requirements on the frequency agents must communicate that is prescribed by the problem structure. The general problem requires every agent to have a model of every other agent in the system. To overcome this, a specific subset of systems, called partially separable, is defined. These systems only require each agent to have a combined summary of the rest of the system. This leads to the definition of an infinitely scalable system, which may contain an infinite number of agents while ensuring the local decisions will converge to the optimal team decision. Examples are given for reconnaissance or information gathering tasks

Asynchronous gradient-based optimisation for team decision making

This paper describes a decentralised asynchronous algorithm for negotiation in team decision and control problems, allowing multiple decision makers to propose and refine future decisions to optimise a common non-linear objective or cost function. A convergence requirement provides an intuitive relationship between the communication frequency, transmission delays and the degree of inter-agent coupling inherent in the system. The coupling is defined by the cross derivative of the objective function. The algorithm is applied to the control of multiple vehicles performing a search task with simulation results given.

Optimising Sensor Layouts for Direct Measurement of Discrete Variables

An optimal sensor layout is attained when a limited number of sensors are placed in an area such that the cost of the placement is minimized while the value of the obtained information is maximized. In this paper, we discuss the optimal sensor layout design problem from first principles, show how an existing optimization criterion (maximum entropy of the measured variables) can be derived, and compare the performance of this criterion with three others that have been reported in the literature for a specific situation for which we have detailed experimental data available. This is achieved by firstly learning a spatial model of the environment using a Bayesian Network, then predicting the expected sensor data in the rest of the space, and finally verifying the predicted results with the experimental measurements. The development of rigorous techniques for optimizing sensor layouts is argued to be an essential requirement for reconfigurable and self-adaptive networks.

Decentralized Decision Making for Multiagent Systems

Decision making in large distributed multiagent systems is a difficult problem. In general, for an agent to make a good decision, it must consider the decisions of all the other agents in the system. This coupling among decision makers has two main causes: (i) the agents share a common objective function (e.g., in a team), or (ii) the agents share constraints (e.g., they must cooperate in sharing a finite resource). The classical approach to this type of problem is to collect all the information from the agents in a single center and solve the resulting optimization problem [see, e.g., Furukawa et al. (2003)]. However, this centralized approach has two main difficulties:

An Indoor Experiment in Decentralized Coordinated Search

This paper addresses the problem of coordinating a team of multiple heterogeneous sensing platforms searching for a single mobile target in a dynamic environment. The proposed implementation of an active sensor network architecture combines a general decentralized Bayesian filtering algorithm with a decentralized coordinated control strategy. In this approach, by communicating with their neighbors on the network, each decision maker builds an equivalent representation of the probability density function of the target state on which they base their control decision.

Bayesian Inferencing on an Aircraft T-Tail Using Probabilistic Surrogates and Uncertainty Quantification

Uncertainty quantification is a notion that has received much interest over the past decade. It involves the extraction of statistical information from a problem with inherent variability. This variability may stem from a lack of knowledge or through observational uncertainty. Traditionally, uncertainty quantification has been a challenging pursuit owing to the lack of efficient methods available. The archetypal uncertainty quantification method is Monte Carlo theory, however, this method possesses a slow convergence rate and is therefore a computational burden in some scenarios. In contrast to Monte Carlo theory, polynomial chaos theory is an alternative approach that offers the ability to estimate statistical moments efficiently. Because polynomial chaos theory behaves like a surrogate model, it is possible to query this inexpensively for information, which allows it to be useful for Bayesian inferencing. This paper builds upon previous work, because a polynomial chaos model is demonstrated to not only ...

Decentralised Decision Making for Ad-hoc Multi-Agent Systems

Decision making in large distributed multi-agent systems is a fundamental problem with a wide range of applications including distributed environmental monitoring, area search and surveillance, and coordination of transportation systems. In general, for an agent to make an good decision, it must consider the decisions of all the other agents in the system. This coupling between decision makers has two main causes: (i) the agents share a common objective function (e.g. they operate as a team), or (ii) the agents have individual goals but share constraints (e.g. they must cooperate in sharing a finite resource). This chapter is focused on the first issue, and assumes the agents are designed to operate as a team. This chapter approaches the multi-agent collaboration problem using distributed optimisation techniques and presents results on the structure of the decision problem and how to exploit sparseness. Although distributed optimisation methods have been studied for over two decades (Baudet in J. ACM 25(2):226–244, 1978; Bertsekas and Tsitsiklis in Parallel and distributed computation: numerical methods. Prentice-Hall, New York, 1989; Automatica 27(1):3–21, 1991; Tsitsiklis et al. in Trans. Autom. Control 31(9):803–812, 1986; Tseng in J. Optim. 1(4):603–619, 1991; Patriksson in J. Comput. Optim. Appl. 9(1):5–42, 1997; Camponogara and Talukdar in IEEE Trans. Syst. Man Cybern. 37(7):732–745, 2007), the existing results and algorithms generally require significant prior configuration that defines the structure of the local interactions and are generally not suitable ad-hoc multi-agent systems.