Models are frequently used to represent complex systems in order to test the systems before they are deployed. Some of the most complicated models are those that represent multi-agent systems (MAS), where there are multiple decision makers. Brahms is an agent-oriented language that models MAS. Three major qualities affect the behavior of these MAS models: workframes that change the state of the system, communication activities that coordinate information between agents, and the schedule of workframes. The primary method to test these models that exists is repeated simulation. Simulation is useful insofar as interesting test cases are used that enable the simulation to explore different behaviors of the model, but simulation alone cannot be fully relied upon to adequately cover the test space, especially in the case of non-deterministic concurrent systems. It takes an exponential number of simulation trials to uncover schedules that reveal unexpected behaviors. This thesis defines a coverage metric to make simulation more meaningful before verification of the model. The coverage metric is divided into three different metrics: workframe coverage, communication coverage, and schedule coverage. Each coverage metric is defined through static analysis of the system, resulting in the coverage requirements of that system. These coverage requirements are compared to the logged output of the simulation run to calculate the coverage of the system. The use of the coverage metric is illustrated in several empirical studies and explored in a detailed case study of the SATS concept (Small Aircraft Transportation System). SATS outlines the procedures aircraft follow around runways that do not have communication towers. The coverage metric quantifies the test effort, and can be used as a basis for future automated test generation and active test.

Debugging multi-agent systems, which are concurrent, distributed, and consist of complex components is difficult, yet crucial. The development of these complex systems is supported by agent-oriented software engineering methodologies which utilise agents as the central design metaphor. The systems that are developed are inherently complex since the components of these systems may interact in flexible and sophisticated ways and traditional debugging techniques are not appropriate. Despite this, very little effort has been applied to developing appropriate debugging tools and techniques. Debugging multi-agent systems without good debugging tools is highly impractical and without suitable debugging support developing and maintaining multi-agent systems will be more difficult than it need be. In this thesis we propose that the debugging process can be supported by following an agent-oriented design methodology, and then using the developed design artifacts in the debugging phase. We propose a domain independent debugging framework which comprises the developed processes and components that are necessary in using design artifacts as debugging artifacts. Our approach is to take a non-formal design artifact, such as an AUML protocol design, and encode it in a machine interpretable manner such that the design can be used as a model of correct system behaviour. These models are used by a run-time debugging system to compare observed behaviour against specified behaviour. We provide details for transforming two design artifact types into equivalent debugging artifacts and show how these can be used to detect bugs. During a debugging episode in which a bug has been identified our debugging approach can provide detailed information about the possible reason for the bug occurring. To determine if this information was useful in helping to debug programs we undertook a thorough empirical study and identified that use of the debugging tool translated to an improvement in debugging performance. We conclude that the debugging techniques developed in this thesis provide effective debugging support for multi-agent systems and by having an extensible framework new design artifacts can be explored and as translations are developed they can be added to the debugging system.

Dans cette thèse nous mettons en avant l'importance de considérer un SMA suivant tous ses aspects : agent, environnement, interaction, organisation et institution. Jusqu'à maintenant, ces aspects ont principalement été traités séparément, et il est difficile d'intégrer des modèles qui portent sur des aspects différents. De plus, nous constatons qu'il existe parfois une ambiguïté sur le rôle de chacun de ces aspects d'un SMA. Il y a donc un besoin de les clarifier et de permettre de tous les intégrer de manière cohérente. Dans cet objectif, nous proposons un cadre conceptuel original pour les SMA qui est fondé sur les quatre quadrants de la vision intégrale de Wilber. Ces quadrants résultent de l'intersection de deux axes d'analyse : l'axe interne – externe et l'axe individuel – collectif. Un intérêt de ce cadre conceptuel est de mettre en évidence le champ d'application de modèles existants, et de faciliter l'intégration de modèles complémentaires. De plus, nous montrons qu'il est adapté à la conception de systèmes ouverts. Enfin nous proposons le méta-modèle MASQ, qui est une formalisation de cette approche conceptuelle. MASQ permet de décrire un SMA à partir de quatre concepts fondamentaux (esprit, objet-corps, espace brut et espace culturel), de relations entre ces concepts, et de lois d'évolution qui déterminent leur dynamique. Ce méta-modèle a pour objectif de mettre en relation des modèles spécifiques complémentaires pour la conception d'un SMA.

In large multiagent games, partial observability, coordination, and credit assignment persistently plague attempts to design good learning algorithms. We provide a simple and efficient algorithm that in part uses a linear system to model the world from a single agent’s limited perspective, and takes advantage of Kalman filtering to allow an agent to construct a good training signal and effectively learn a near-optimal policy in a wide variety of settings. A sequence of increasingly complex empirical tests verifies the efficacy of this technique. / Singapore-MIT Alliance (SMA)

Research in mobile ad-hoc networks has focused on situations in which nodes have no control over their movements. We investigate an important but overlooked domain in which nodes do have control over their movements. Reinforcement learning methods can be used to control both packet routing decisions and node mobility, dramatically improving the connectivity of the network. We first motivate the problem by presenting theoretical bounds for the connectivity improvement of partially mobile networks and then present superior empirical results under a variety of different scenarios in which the mobile nodes in our ad-hoc network are embedded with adaptive routing policies and learned movement policies.

In this thesis I present a language for instructing a sheet of identically-programmed, flexible, autonomous agents (``cells'') to assemble themselves into a predetermined global shape, using local interactions. The global shape is described as a folding construction on a continuous sheet, using a set of axioms from paper-folding (origami). I provide a means of automatically deriving the cell program, executed by all cells, from the global shape description. With this language, a wide variety of global shapes and patterns can be synthesized, using only local interactions between identically-programmed cells. Examples include flat layered shapes, all plane Euclidean constructions, and a variety of tessellation patterns. In contrast to approaches based on cellular automata or evolution, the cell program is directly derived from the global shape description and is composed from a small number of biologically-inspired primitives: gradients, neighborhood query, polarity inversion, cell-to-cell contact and flexible folding. The cell programs are robust, without relying on regular cell placement, global coordinates, or synchronous operation and can tolerate a small amount of random cell death. I show that an average cell neighborhood of 15 is sufficient to reliably self-assemble complex shapes and geometric patterns on randomly distributed cells. The language provides many insights into the relationship between local and global descriptions of behavior, such as the advantage of constructive languages, mechanisms for achieving global robustness, and mechanisms for achieving scale-independent shapes from a single cell program. The language suggests a mechanism by which many related shapes can be created by the same cell program, in the manner of D'Arcy Thompson's famous coordinate transformations. The thesis illuminates how complex morphology and pattern can emerge from local interactions, and how one can engineer robust self-assembly.

Most Artificial Intelligence (AI) work can be characterized as either ``high-level'' (e.g., logical, symbolic) or ``low-level'' (e.g., connectionist networks, behavior-based robotics). Each approach suffers from particular drawbacks. High-level AI uses abstractions that often have no relation to the way real, biological brains work. Low-level AI, on the other hand, tends to lack the powerful abstractions that are needed to express complex structures and relationships. I have tried to combine the best features of both approaches, by building a set of programming abstractions defined in terms of simple, biologically plausible components. At the ``ground level'', I define a primitive, perceptron-like computational unit. I then show how more abstract computational units may be implemented in terms of the primitive units, and show the utility of the abstract units in sample networks. The new units make it possible to build networks using concepts such as long-term memories, short-term memories, and frames. As a demonstration of these abstractions, I have implemented a simulator for ``creatures'' controlled by a network of abstract units. The creatures exist in a simple 2D world, and exhibit behaviors such as catching mobile prey and sorting colored blocks into matching boxes. This program demonstrates that it is possible to build systems that can interact effectively with a dynamic physical environment, yet use symbolic representations to control aspects of their behavior.

The electric power systems in U.S. Navy ships supply energy to sophisticated systems for weapons, communications, navigation and operation. The reliability and survivability of the Shipboard Power System (SPS) are critical to the mission of a surface combatant ship, especially under battle conditions. In the event of battle, various weapons might attack a ship. When a weapon hits the ship it can cause severe damage to the electrical system on the ship. This damage can lead to de-energization of critical loads on a ship that can eventually decrease a ship’s ability to survive the attack. It is very important, therefore, to maintain availability of energy to the connected loads that keep the power systems operational. Technology exists that enables the detection of an incoming weapon and prediction of the geographic area where the incoming weapon will hit the ship. This information can then be used to take reconfiguration actions before the actual hit so that the actual damage caused by the weapon hit is reduced. The Power System Automation Lab (PSAL) has proposed a unique concept called "Predictive Reconfiguration" which refers to performing reconfiguration of a ship’s power system before a weapon hit to reduce the potential damage to the electrical system caused by the impending weapon hit. The concept also includes reconfiguring the electrical system to restore power to as much of the healthy system as possible after the weapon hit. This dissertation presents a new methodology for Predictive Reconfiguration of a Shipboard Power System (SPS). This probabilistic approach includes a method to assess the damage that will be caused by a weapon hit. This method calculates the expected probability of damage for each electrical component on the ship. Also a heuristic method is included, which uses the expected probability of damage to determine reconfiguration steps to reconfigure the ship’s electrical network to reduce the damage caused by a weapon hit. This dissertation also presents a modified approach for performing a reconfiguration for restoration after the weapon hits the system. In this modified approach, an expert system based restoration method restores power to loads de-energized due to the weapon hit. These de-energized loads are restored in a priority order. The methods were implemented using multi-agent technology. A test SPS model based on the electrical layout of a non-nuclear surface combatant ship was presented. Complex scenarios representing electrical casualties caused due to a weapon hit, on the test SPS model, were presented. The results of the Predictive Reconfiguration methodology for complex scenarios were presented to illustrate the effectiveness of the developed methodology.

We investigate the use of self-assembly in a robotic system as a means of responding to dierent environmental contingencies. Self-assembly is the mechanism through which agents in a multi-robot system autonomously form connections with one another to create larger composite robotic entities. Initially, we consider a simple response mechanism that uses stochastic self-assembly without any explicit control over the resulting morphology | the robots self-assemble into a larger, randomly shaped composite entity if the task they encounter is beyond the physical capabilities of a single robot. We present distributed behavioural control that enables a group of robots to make this collective decision about when and if to self-assemble in the context of a hill crossing task. In a series of real-world experiments, we analyse the eect of dierent distributed timing and decision strategies on system performance. Outside of a task execution context, we present fully decentralised behavioural control capable of creating periodically repeating global morphologies. We then show how arbitrary morphologies can be generated by abstracting our behavioural control into a morphology control language and adding symbolic communication between connected agents. Finally, we integrate our earlier distributed response mechanism into the morphology control language. We run simulated and real-world experiments to demonstrate a self-assembling robotic system that can respond to varying environmental contingencies by forming dierent appropriate morphologies.

Simulating large numbers of agents, performing complex behaviors in realistic environments is a difficult problem with applications in robotics, computer graphics and animation. A multi-agent system can be a useful tool for studying a range of situations in simulation in order to plan and train for actual events. Systems supporting such simulations can be used to study and train for emergency or disaster scenarios including search and rescue, civilian crowd control, evacuation of a building, and many other training situations. This work describes our approach to multi-agent systems which integrates a roadmap-based approach with agent-based systems for groups of agents performing a wide range of behaviors. The system that we have developed is highly customizable and allows us to study a variety of behaviors and scenarios. The system is tunable in the kinds of agents that can exist and parameters that describe the agents. The agents can have any number of behaviors which dictate how they react throughout a simulation. Aspects that are unique to our approach to multi-agent group behavior are the environmental encoding that the agents use when navigating and the extensive usage of the roadmap in our behavioral framework. Our roadmap-based approach can be utilized to encode both basic and very complex environments which include multi- level buildings, terrains and stadiums. In this work, we develop techniques to improve the simulation of multi-agent systems. The movement strategies we have developed can be used to validate agent movement in a simulated environment and evaluate building designs by varying portions of the environment to see the effect on pedestrian flow. The strategies we develop for searching and tracking improve the ability of agents within our roadmap-based framework to clear areas and track agents in realistic environments. The application focus of this work is on pursuit-evasion and evacuation planning. In pursuit-evasion, one group of agents, the pursuers, attempts to find and capture another set of agents, the evaders. The evaders have a goal of avoiding the pursuers. In evacuation planning, the evacuating agents attempt to find valid paths through potentially complex environments to a safe goal location determined by their environmental knowledge. Another group of agents, the directors may attempt to guide the evacuating agents. These applications require the behaviors created to be tunable to a range of scenarios so they can reflect real-world reactions by agents. They also potentially require interaction and coordination between agents in order to improve the realism of the scenario being studied. These applications illustrate the scalability of our system in terms of the number of agents that can be supported, the kinds of realistic environments that can be handled, and behaviors that can be simulated.