This paper describes an exercise in building a complete robot aimed at being as small as possible but using off-the-shelf components exclusively. The result is an autonomous mobile robot slightly larger than one cubic inch which incorporates sensing, actuation, onboard computation, and onboard power supplies. Nicknamed Squirt, this robot acts as a 'bug', hiding in dark corners and venturing out in the direction of last heard noises, only moving after the noises are long gone.

This Ph.D. thesis presents the design and implementation of a biologically inspired four-phase walking strategy using behaviours for a four legged walking robot. In particular, the walking strategy addresses the balance issue, including both static and dynamic balance that were triggered non-deterministically based on the robot’s realtime interaction with the environment. Four parallel Subsumption Architectures (SA) and a simple Central Pattern Producer (CPP) are employed in the physical implementation of the walking strategy. An implementation framework for such a parallel Subsumption Architecture is also proposed to facilitate the reusability of the system. A Reinforcement Learning (RL) method was integrated into the CPP to allow the robot to learn the optimal walking cycle interval (OWCI), appropriate for the robot walking on various terrain conditions. Experimental results demonstrate that the robot employs the proposed walking strategy and can successfully carry out its walking behaviours under various experimental terrain conditions, such as flat ground, incline, decline and uneven ground. Interactions of all the behaviours of the robot enable it to exhibit a combination of both preset and emergent walking behaviours.

Conventional method of development which repeats phases of design, implementation and testing is not adequate for systems like intelligent agents for which complex behavior is required but specification is unclear at the beginning of development process. Incremental design of agent's model during simulation seems more suitable for it enables direct feedback in behavioral changes of a system. This interactivity speeds up development process and helps to uncover parts of a space of all models to designer - thanks to new knowledge acquired during simulation. This thesis aims to provide comparison of suitability of two agent architectures and respective platforms in context of SmallDEVS modeling and simulation framework for this methodology of development. First approach is reactive and decentralized subsumption architecture based on DEVS formalism and the second one is PNagent, framework realizing deliberative BDI architecture using Object Oriented Petri Nets.

Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - CAPES / A very important aspect in robotics is the decision making and execution the system uses to achieve its goals. In literature, many different approaches can be found about how the robot must behave in different situations in order to have a more robust system. Subsumption architecture is one of the most used and referenced in the area. In this architecture, the global task is divided into subtasks which are performed by behaviors organized in hierarchical layers. However, little research has been done regarding the stability analysis of this architecture. Behavioral changes imply in controller switching, which can lead the system to instability even in cases where all controllers are stable. In this work, a subsumption architecture with guaranteed stability is presented based on the theory of switched systems with main control objective. In addition, a formalism capable of allowing behaviors modeling in a simple and fast way is proposed based on the theory of discrete events systems. Tests in real environments were performed with the Pioneer P3-DX robot and obtained results demonstrate the proposed approach effectiveness. / Um aspecto muito importante na robótica é a tomada de decisão e execução que o sistema utiliza para alcançar seus objetivos. Na literatura, existem vários trabalhos diferentes para abordar como o robô deve se comportar diante de várias situações diferentes a fim de trazer uma maior robustez ao sistema, sendo a arquitetura de subsunção uma das mais utilizadas e referenciadas na área. Nesta arquitetura, a tarefa global é dividida em subtarefas que são executadas por comportamentos organizados em camadas de forma hierárquica. No entanto, pouco se pesquisa no que diz respeito a análise de estabilidade desta arquitetura, sendo que as mudanças de comportamento implicam em chaveamento de controladores, que por sua vez podem levar o sistema a instabilidade mesmo em casos em que todos os controladores sejam estáveis. Desta forma, neste trabalho é apresentada uma arquitetura de subsunção com prova de estabilidade garantida com base na teoria de controle chaveado com objetivo de controle principal. Além disso, um formalismo capaz de permitir a modelagem dos comportamentos de forma simples e rápida é proposto com base na teoria de sistemas a eventos discretos. Testes em ambientes reais foram realizados com o robô Pioneer P3-DX e os resultados obtidos comprovam a eficácia da abordagem proposta.