Computer Science

The decreased size and cost of Unmanned Aerial Vehicles (UAVs) and Unmanned Ground Vehicles (UGVs) has enabled the use of swarms of unmanned autonomous vehicles to accomplish a variety of tasks. By utilizing swarming behaviors, it is possible to efficiently accomplish coordinated tasks while minimizing per-drone computational requirements. Some drones rely on decentralized protocols that exhibit emergent behavior across the swarm. While fully decentralized algorithms remove obvious attack vectors their susceptibility to external influence is less understood. This work investigates the influences that can compromise the functionality of an autonomous swarm leading to hazardous situations and cascading vulnerabilities. When a swarm is tasked with missions involving the safety or health of humans, external influences could have serious consequences. The adversarial swarm in this work utilizes an attack vector embedded within the decentralized movement algorithm of a previously defined autonomous swarm designed to create a perimeter sentry swarm. Various simulations confirm the adversarial swarm's ability to capture significant portions (6-23%) of the perimeter.

This paper presents a novel virus propagation model using NetLogo. The model allows agents to move across multiple sites using different routes. Routes can be configured, enabled for mobility and (un)locked down independently. Similarly, locations can also be (un)locked down independently. Agents can get infected, propagate their infections to others, can take precautions against infection and also subsequently recover from infection. This model contains certain features that are not present in existing models. The model may be used for educational and research purposes, and the code is made available as open source. This model may also provide a broader framework for more detailed simulations. The results presented are only to demonstrate the model functionalities and do not serve any other purpose.

In this paper, we present the strategy of Agent Madoff, which is a heuristic-based negotiation agent that won 2nd place at the Automated Negotiating Agents Competition (ANAC 2017). Agent Madoff is implemented to play the game Diplomacy, which is a strategic board game that mimics the situation during World War I. Each player represents a major European power which has to negotiate with other forces and win possession of a majority supply centers on the map. We propose a design architecture which consists of 3 components: heuristic module, acceptance strategy and bidding strategy. The heuristic module, responsible for evaluating which regions on the graph are more worthy, considers the type of region and the number of supply centers adjacent to the region and return a utility value for each region on the map. The acceptance strategy is done on a case-by-case basis according to the type of the order by calculating the acceptance probability using a composite function. The bidding strategy adopts a defensive approach aimed to neutralize attacks and resolve conflict moves with other players to minimize our loss on supply centers.

In this paper we explore how actor-critic methods in deep reinforcement learning, in particular Asynchronous Advantage Actor-Critic (A3C), can be extended with agent modeling. Inspired by recent works on representation learning and multiagent deep reinforcement learning, we propose two architectures to perform agent modeling: the first one based on parameter sharing, and the second one based on agent policy features. Both architectures aim to learn other agents' policies as auxiliary tasks, besides the standard actor (policy) and critic (values). We performed experiments in both cooperative and competitive domains. The former is a problem of coordinated multiagent object transportation and the latter is a two-player mini version of the Pommerman game. Our results show that the proposed architectures stabilize learning and outperform the standard A3C architecture when learning a best response in terms of expected rewards.

Autonomous agents are seen as a prominent technology to be applied in industrial scenarios. Classical automation solutions are struggling with challenges related to high dynamism, prompt actuation, heterogeneous entities, including humans, and decentralised decision-making. Besides promoting concepts, languages, and tools to face such challenges, agents must also provide high reliability. To assess how appropriate and mature are agents for industrial applications, we have investigated its application in two scenarios of the gas and oil industry. This paper presents the development of systems and the initial results highlighting the advantages and drawbacks of the agents approach when compared with the existing automation solutions.

This paper presents algorithms of decision making agents for an integrated air defense (IAD) system. The advantage of using agent based over conventional decision making system is its ability to automatically detect and track targets and if required allocate weapons to neutralize threat in an integrated mode. Such approach is particularly useful for futuristic network centric warfare. Two agents are presented here that perform the basic decisions making tasks of command and control (C2) like detection and action against jamming, threat assessment and weapons allocation, etc. The belief-desire-intension (BDI) architectures stay behind the building blocks of these agents. These agents decide their actions by meta level plan reasoning process. The proposed agent based IAD system runs without any manual inputs, and represents a state of art model for C2 autonomy.

Modelling and computational methods have been essential in advancing quantitative science, especially in the past two decades with the availability of vast amount of complex, voluminous, and heterogeneous data. In particular, there has been a surge of interest in agent-based modelling, largely due to its capabilities to exploit such data and make significant projections. However, any well-established quantitative method relies on theoretical frameworks for both construction and analysis. While the computational aspects of agent-based modelling have been detailed in existing literature, the underlying theoretical basis has rarely been used in its construction. In this exposition, we provide an overview of the theoretical foundation of agent-based modelling and establish a relationship with its computational implementation. In addition to detailing the main characteristics of this computational methodology, we illustrate its application to simulating the spread of an infectious disease in a simple, dynamical process. As the use of agent-based models expands to various disciplines, our review highlights the need for directed research efforts to develop theoretical methods and analytical tools for the analysis of such models.

The premise of this working paper is based around agent-based simulation models and how to go about creating them from given incomplete information. Agent-based simulations are stochastic simulations that revolve around groups of agents that each have their own characteristics and can make decisions. Such simulations can be used to emulate real life situations and to create hypothetical situations without the need for real-world testing prior. Here we describe the development of an agent-based simulation model for studying future digital mental health scenarios. An incomplete conceptual model has been used as the basis for this development. To define differences in responses to stimuli we employed fuzzy decision making logic. The model has been implemented but not been used for structured experimentation yet. This is planned as our next step.

Simulation models of pedestrian dynamics have become an invaluable tool for evacuation planning. Typically crowds are assumed to stream unidirectionally towards a safe area. Simulated agents avoid collisions through mechanisms that belong to each individual, such as being repelled from each other by imaginary forces. But classic locomotion models fail when collective cooperation is called for, notably when an agent, say a first-aid attendant, needs to forge a path through a densely packed group. We present a controlled experiment to observe what happens when humans pass through a dense static crowd. We formulate and test hypothesis on salient phenomena. We discuss our observations in a psychological framework. We derive a model that incorporates: agents' perception and cognitive processing of a situation that needs cooperation; selection from a portfolio of behaviours, such as being cooperative; and a suitable action, such as swapping places. Agents' ability to successfully get through a dense crowd emerges as an effect of the psychological model.

The COVID-19 pandemic has affected travel behaviors and transportation system operations, and cities are grappling with what policies can be effective for a phased reopening shaped by social distancing. This edition of the white paper updates travel trends and highlights an agent-based simulation model's results to predict the impact of proposed phased reopening strategies. It also introduces a real-time video processing method to measure social distancing through cameras on city streets.