Submitted by Lucienne Costa (lucienneferreira@ufcg.edu.br) on 2018-05-03T18:38:37Z No. of bitstreams: 1 WILLY CARVALHO TIENGO – TESE (PPGCC) 2018.pdf: 4153575 bytes, checksum: 929b905dca8b61fcb0f831264752540f (MD5) / Made available in DSpace on 2018-05-03T18:38:37Z (GMT). No. of bitstreams: 1 WILLY CARVALHO TIENGO – TESE (PPGCC) 2018.pdf: 4153575 bytes, checksum: 929b905dca8b61fcb0f831264752540f (MD5) Previous issue date: 2018 / No Brasil, o transporte rodoviário é responsável por 58% do transporte de carga, que tem os acidentes como um grande problema, pois, em geral, esses ocasionam muitas vítimas, prejuízos econômicos relevantes e em alguns casos danos ambientais decorrentes de derramamento de carga. Estudos apontam que os prejuízos com os acidentes no transporte de carga em 2012 foram de mais de 9 bilhões de reais. Estudo realizado em 2007 pela PAMCARY, corretora de seguros e gestora de riscos, revelou que os eventos que combinam maior frequência e gravidade são tombamento e capotagem. Nesse sentido, esta pesquisa consiste na elaboração de um assistente avançado para motorista que objetiva alertar previamente sobre a velocidade limite da curva, a fim de diminuir os riscos de tombamento. Em outras palavras, consiste em buscar mitigar o problema auxiliando o motorista para que ele mantenha o veículo em uma velocidade segura, por meio de alertas e em prazo adequado, que permitam ao motorista tomar medidas corretivas em caso de estado inseguro. A solução foi desenvolvida a partir de uma arquitetura modular, que funciona da seguinte forma: por meio de sensores (velocidade, GPS e posição do acelerador), associado a mapas digitais, o risco de acidente é controlado constantemente. Com isso, um dispositivo poderia ser embarcado na cabine do veículo para emitir alertas visual e auditivo de risco de tombamento. A solução utiliza o indicador de estabilidade chamado Limiar Estático de Tombamento que, associado à informação a priori de mapas digitais, permite o cálculo do risco de tombamento com diferentes abordagens. No contexto da pesquisa, foram desenvolvidas 04 versões de assistentes. Além disso, foi proposto um arcabouço de simulação microscópica de trânsito baseado no modelo de raciocínio prático denominado de belief-desire-intention (BDI) para permitir o desenvolvimento e a validação de agentes inteligentes para Sistemas Avançados de Assistência ao Motorista de maneira rápida, flexível e fácil. Para avaliar o potencial dos assistentes, foi escolhida a BR-101, estrada federal de Alagoas com mais ocorrências de tombamento. Nessa rodovia, foram simulados 400 veículos para avaliar o desempenho dos assistentes propostos. Em particular, foram investigadas a efetividade, intrusividade, omissão e a segurança para avaliar o desempenho dos assistentes. / In Brazil, highway transportation is responsible for 58% of cargo transport. A relevant problem associated to cargo transport are the accidents, that generally cause an elevated number of victims, relevant economic losses and, in some cases, damages to the environment due to cargo spills, since there are also dangerous products being transported. Researches point out that the cost of accidents in cargo transportation in 2012 was more than BRL 9 billion. A study performed in 2007 by PAMCARY revealed the accidents profile: the events that combine higher frequency and gravity are rollover and tipping (considered here as the same nature). In this study, incompatible speed and fatigue, factors that are related to human actions, were pointed out as main causes of accidents; for another hand, sharp curve and poorly maintained roads are contributing factors to accidents. Therefore, the research proposal consists of the adoption of an assistant for warning in advance of over speed for a specific curve. This may reduce rollover risks. In other words, it would be mitigated the problem by helping the driver to maintain the vehicle in a safe speed, through customized alerts just in time to allow the driver to take corrective maneuvers in case of unsafe state. The solution is a modular architecture, which works as follows: through sensors (speed, GPS and throttle position) associated with digital maps, it is controlled the risk of accident constantly. With that, an embedded device at the vehicle’s cab could to emit visual and sound alerts warning the risk of rollover. In this work, it is proposed the adoption of the stability indicator known as Static Rollover Threshold, which is combined with a priori information from digital maps to allow the calculation of the rollover risk by different approaches. In the context of this research, 04 versions of assistants were developed. In addition, a microscopic traffic simulation framework was proposed based on the practical reasoning model named belief-desire-intention (BDI) to support the development and validation of intelligent agents for Advanced Driver Assistance Systems in a fast, flexible and easy way. To evaluate the assistants’ potential, the BR-101, Federal Highway of Alagoas with more occurrence of rollover, was chosen. On this highway, 400 vehicles were simulated to evaluate the performance of the proposed assistants. The effectiveness, intrusiveness, omission and safety of the assistants were investigated.

In this thesis we extend BDI logics, which are normal multimodal logics with an arbitrary set of normal modal operators, from three different perspectives. Firstly, based on some recent developments in modal logic, we examine BDI logics from a combining logic perspective and apply combination techniques like fibring/dovetailing for explaining them. The second perspective is to extend the underlying logics so as to include action constructs in an explicit way based on some recent action-related theories. The third perspective is to adopt a non-monotonic logic like defeasible logic to reason about intentions in BDI. As such, the research captured in this thesis is theoretical in nature and situated at the crossroads of various disciplines relevant to Artificial Intelligence (AI). More specifically this thesis makes the following contributions: 1. Combining BDI Logics through fibring/dovetailing: BDI systems modeling rational agents have a combined system of logics of belief, time and intention which in turn are basically combinations of well understood modal logics. The idea behind combining logics is to develop general techniques that allow to produce combinations of existing and well understood logics. To this end we adopt Gabbay's fibring/dovetailing technique to provide a general framework for the combinations of BDI logics. We show that the existing BDI framework is a dovetailed system. Further we give conditions on the fibring function to accommodate interaction axioms of the type G [superscript k,l,m,n] ([diamond][superscript k] [superscript l] [phi] [implies] [superscript m] [diamond][superscript n] [phi]) based on Catach's multimodal semantics. This is a major result when compared with other combining techniques like fusion which fails to accommodate axioms of the above type. 2. Extending the BDI framework to accommodate Composite Actions: Taking motivation from a recent work on BDI theory, we incorporate the notion of composite actions, [pi]-1; [pi]-2 (interpreted as [pi]-1 followed by [pi]-2), to the existing BDI framework. To this end we introduce two new constructs Result and Opportunity which helps in reasoning about the actual execution of such actions. We give a set of axioms that can accommodate the new constructs and analyse the set of commitment axioms as given in the original work in the background of the new framework. 3. Intention reasoning as Defeasible reasoning: We argue for a non-monotonic logic of intention in BDI as opposed to the usual normal modal logic one. Our argument is based on Bratman's policy-based intention. We show that policy-based intention has a defeasible/non-monotonic nature and hence the traditional normal modal logic approach to reason about such intentions fails. We give a formalisation of policy-based intention in the background of defeasible logic. The problem of logical omniscience which usually accompanies normal modal logics is avoided to a great extend through such an approach.

In recent years, multi-agent traffic simulation has been widely used to accurately evaluate the performance of a road network considering individual and dynamic movements of vehicles under a virtual roadway environment. Given initial traffic demands and road conditions, the simulation is executed with multiple iterations and provides users with converged roadway conditions for the performance evaluation. For an accurate traffic simulation model, the driver's learning behavior is one of the major components to be concerned, as it affects road conditions (e.g., traffic flows) at each iteration as well as performance (e.g., accuracy and computational efficiency) of the traffic simulation. The goal of this study is to propose a realistic learning behavior model of drivers concerning their uncertain perception and interactions with other drivers. The proposed learning model is based on the Extended Belief-Desire-Intention (E-BDI) framework and two major decisions arising in the field of transportation (i.e., route planning and decision-making at an intersection). More specifically, the learning behavior is modeled via a dynamic evolution of a Bayesian network (BN) structure. The proposed dynamic learning approach considers three underlying assumptions: 1) the limited memory of a driver, 2) learning with incomplete observations on the road conditions, and 3) non-stationary road conditions. Thus, the dynamic learning approach allows driver agents to understand real-time road conditions and estimate future road conditions based on their past knowledge. In addition, interaction behaviors are also incorporated in the E-BDI framework to address influences of interactions on the driver's learning behavior. In this dissertation work, five major human interactions adopted from a social science literature are considered: 1) accommodation, 2) collaboration, 3) compromise, 4) avoidance, and 5) competition. The first three interaction types help to mimic information exchange behaviors between drivers (e.g., finding a route using a navigation system) while the last two interaction types are relevant with behaviors involving non-information exchange behaviors (e.g., finding a route based on a driver's own experiences). To calibrate the proposed learning behavior model and evaluate its performance in terms of inference accuracy and computational efficiency, drivers' decision data at intersections are collected via a human-in-the-loop experiment involving a driving simulator. Moreover, the proposed model is used to test and demonstrate the impact of five interactions on drivers' learning behavior under an en route planning scenario with real traffic data of Albany, New York, and Phoenix, Arizona. In this dissertation work, two major traffic simulation platforms, AnyLogic® and DynusT®, are used for the demonstration purposes. The experimental results reveal that the proposed model is effective in modeling realistic learning behaviors of drivers in conduction with interactions with other drivers.

The BDI agent architecture includes a plan library containing pre-de ned plans. The plan library is included in the agent architecture to reduce the need for expensive means-end reasoning, however can hinder the agent's e ectiveness when operating in a changing environment. Existing research on integrating di erent planning methods into the BDI agent to overcome this limitation include HTNs, state-space planning and Graphplan. Genetic Algorithms (GAs) have not yet been used for this purpose. This dissertation investigates the feasibility of using GAs as a plan modi cation mechanism for BDI agents. It covers the design of a plan structure that can be encoded into a binary string, which can be operated on by the genetic operators. The e ectiveness of the agent in a changing environment is compared to an agent without the GA plan modification mechanism. The dissertation shows that GAs are a feasible plan modification mechanism for BDI agents. / Information Science

The BDI agent architecture includes a plan library containing pre-defined plans. The plan library is included in the agent architecture to reduce the need for expensive means-end reasoning, however can hinder the agent’s effectiveness when operating in a changing environment. Existing research on integrating different planning methods into the BDI agent to overcome this limitation include HTNs, state-space planning and Graphplan. Genetic Algorithms (GAs) have not yet been used for this purpose. This dissertation investigates the feasibility of using GAs as a plan modification mechanism for BDI agents. It covers the design of a plan structure that can be encoded into a binary string, which can be operated on by the genetic operators. The effectiveness of the agent in a changing environment is compared to an agent without the GA plan modification mechanism. The dissertation shows that GAs are a feasible plan modification mechanism for BDI agents. / Information Science

Målet med den här datorbaserade filosofiska utredningen inom kognitionsvetenskap är att utforska intentionalitet i kollektiva beteenden hos artificiella svärmar. Två definitioner av intentionalitet utforskades; som representationer hos agenter och som observerbara attribut hos agenter, även kallat intentional stance. För den representativa definitionen användes en modell av kollektiv intentionalitet som integrerar två olika ståndpunkter, singularståndpunkten och pluralståndpunkten av kollektiv intentionalitet. Modellen har fem villkor för intentionalitet enligt SharedPlans. Genom att använda Belief-Desire-Intention-modellen för intelligenta agenter operationaliserades villkoren till möjliga representationer. En implementation av en målinriktad artificiell svärm i NetLogo analyserades genom att studera hur väl den uppfyllde de operationaliserade villkoren. Fyra av fem villkor var uppfyllda. Flera simuleringar med olika hastighet genomfördes även under observation. Dessa visade att processen kunde delas upp i tre faser med olika egenskaper. Den utforskande fasen hade gemensam intentionalitet centrerad till ett fåtal aktiva individer. Beslutsfasen hade individuella intentioner som kunde stå i konflikt med varandra medan gemensamma intentioner strävade mot samma mål. I flyttfasen var de individuella intentionerna att förhålla sig till varandra, vilket fick gruppen att upplevas som en enhet med intentionen att flytta gruppen. Resultaten visade att intentionalitet kan observeras och analyseras hos den här artificiella svärmen. Däremot har svärmen inte kollektiv intentionalitet utifrån båda ståndpunkterna.

Studying and simulating social systems including human groups and societies can be a complex problem. In order to build a model that simulates humans' actions, it is necessary to consider the major factors that affect human behavior. Norms are one of these factors: social norms are the customary rules that govern behavior in groups and societies. Norms are everywhere around us, from the way people handshake or bow to the clothes they wear. They play a large role in determining our behaviors. Studies on norms are much older than the age of computer science, since normative studies have been a classic topic in sociology, psychology, philosophy and law. Various theories have been put forth about the functioning of social norms. Although an extensive amount of research on norms has been performed during the recent years, there remains a significant gap between current models and models that can explain real-world normative behaviors. Most of the existing work on norms focuses on abstract applications, and very few realistic normative simulations of human societies can be found. The contributions of this dissertation include the following: 1) a new hybrid technique based on agent-based modeling and Markov Chain Monte Carlo is introduced. This method is used to prepare a smoking case study for applying normative models. 2) This hybrid technique is described using category theory, which is a mathematical theory focusing on relations rather than objects. 3) The relationship between norm emergence in social networks and the theory of tipping points is studied. 4) A new lightweight normative architecture for studying smoking cessation trends is introduced. This architecture is then extended to a more general normative framework that can be used to model real-world normative behaviors. The final normative architecture considers cognitive and social aspects of norm formation in human societies. Normative architectures based on only one of these two aspects exist in the literature, but a normative architecture that effectively includes both of these two is missing.

This dissertation investigates high-level decision making for agents that are both goal and utility driven. We develop a partially observable Markov decision process (POMDP) planner which is an extension of an agent programming language called DTGolog, itself an extension of the Golog language. Golog is based on a logic for reasoning about action—the situation calculus. A POMDP planner on its own cannot cope well with dynamically changing environments and complicated goals. This is exactly a strength of the belief-desire-intention (BDI) model: BDI theory has been developed to design agents that can select goals intelligently, dynamically abandon and adopt new goals, and yet commit to intentions for achieving goals. The contribution of this research is twofold: (1) developing a relational POMDP planner for cognitive robotics, (2) specifying a preliminary BDI architecture that can deal with stochasticity in action and perception, by employing the planner. / Computing / M. Sc. (Computer Science)

This dissertation investigates high-level decision making for agents that are both goal and utility driven. We develop a partially observable Markov decision process (POMDP) planner which is an extension of an agent programming language called DTGolog, itself an extension of the Golog language. Golog is based on a logic for reasoning about action—the situation calculus. A POMDP planner on its own cannot cope well with dynamically changing environments and complicated goals. This is exactly a strength of the belief-desire-intention (BDI) model: BDI theory has been developed to design agents that can select goals intelligently, dynamically abandon and adopt new goals, and yet commit to intentions for achieving goals. The contribution of this research is twofold: (1) developing a relational POMDP planner for cognitive robotics, (2) specifying a preliminary BDI architecture that can deal with stochasticity in action and perception, by employing the planner. / Computing / M. Sc. (Computer Science)