No description available.

As robotic systems move from well-controlled settings to increasingly unstructured environments, they are required to operate in highly dynamic and cluttered scenarios. Finding an object, estimating its pose, and tracking its pose over time within such scenarios are challenging problems. Although various approaches have been developed to tackle these problems, the scope of objects addressed and the robustness of solutions remain limited. In this thesis, we target a robust object perception using visual sensory information, which spans from the traditional monocular camera to the more recently emerged RGB-D sensor, in unstructured environments. Toward this goal, we address four critical challenges to robust 6-DOF object pose estimation and tracking that current state-of-the-art approaches have, as yet, failed to solve. The first challenge is how to increase the scope of objects by allowing visual perception to handle both textured and textureless objects. A large number of 3D object models are widely available in online object model databases, and these object models provide significant prior information including geometric shapes and photometric appearances. We note that using both geometric and photometric attributes available from these models enables us to handle both textured and textureless objects. This thesis presents our efforts to broaden the spectrum of objects to be handled by combining geometric and photometric features. The second challenge is how to dependably estimate and track the pose of an object despite the clutter in backgrounds. Difficulties in object perception rise with the degree of clutter. Background clutter is likely to lead to false measurements, and false measurements tend to result in inaccurate pose estimates. To tackle significant clutter in backgrounds, we present two multiple pose hypotheses frameworks: a particle filtering framework for tracking and a voting framework for pose estimation. Handling of object discontinuities during tracking, such as severe occlusions, disappearances, and blurring, presents another important challenge. In an ideal scenario, a tracked object is visible throughout the entirety of tracking. However, when an object happens to be occluded by other objects or disappears due to the motions of the object or the camera, difficulties ensue. Because the continuous tracking of an object is critical to robotic manipulation, we propose to devise a method to measure tracking quality and to re-initialize tracking as necessary. The final challenge we address is performing these tasks within real-time constraints. Our particle filtering and voting frameworks, while time-consuming, are composed of repetitive, simple and independent computations. Inspired by that observation, we propose to run massively parallelized frameworks on a GPU for those robotic perception tasks which must operate within strict time constraints.

The phenomenon of aggression has been of interest to psychologists for many years, and has resulted in a variety of theories which-attempt to explain its existence in man. Aggression is prevalent in our primary schools today and it was this observation which initiated the research project. The Relationship Theory was applied in order to gain insight into the life-world of the aggressive child and to explore possible causes which may originate from changes in our society. / Psychology of Education / M. Ed. with specialisation in Guidance and Counselling (Psychology of Education)

A contribuição e objetivo desta tese é desenvolver um modelo de redes neurais artificiais, baseado em princípios de memória associativa, capaz de resolver o problema de planejamento e controle ponto a ponto de trajetória de um braço mecânico imerso em um ambiente parcialmente conhecido e/ou sujeito a ruídos. O modelo proposto é formado por dois planos: plano seqüência temporal e plano ângulo. Para o plano seqüência temporal, o novo modelo proposto chamado de Memória Associativa Multidirecional Temporal (TMAM) é capaz de armazenar e recuperar n-tuplas de informações, lidar com informações ruidosas e/ou incompletas e aprender seqüências temporais. TMAM utiliza representação contínua e realimentação autoassociativa. O plano ângulo é formado pelo modelo RBF que é responsável por produzir as informações de ângulos das juntas do braço mecânico. A composição dos dois planos forma o sistema completo que é responsável pelo planejamento e controle ponto a ponto de trajetória. Em resumo, o sistema recebe informações do ponto origem e do ponto alvo, estabelece uma trajetória para atingir o ponto alvo a partir do ponto de origem e transforma os pontos espaciais da trajetória em valores de ângulos das juntas. Os resultados obtidos mostram que o modelo TMAM é capaz de recuperar, interpelar e extrapolar pontos nas seqüências, é capaz de gerar trajetórias, de memorizar seqüências de diferentes tamanhos e de lidar com duas trajetórias ao mesmo tempo. O modelo apresenta também rápido treinamento. O modelo RBF é capaz de recuperar as saídas desejadas apresentando um erro pequeno e é capaz de receber um padrão que apresenta um ponto final inatingível e gerar um conjunto de ângulos que representa um ponto final atingível. / The aim of this project is to develop an artificial neural networks model based on principles of associative memory. This neural network model must be able to solve the problem of trajectory planning and point to point control of a robot arm, which is located in a partially known and/or noisy environment. The proposed model is composed by two surfaces: the temporal sequence surface and the angle surface. For the temporal sequence surface the new propose model Temporal Multidirectional Associative Memmy (TMAM) is able to store and recall n-tuplas of information, to deal with noisy and/or incomplete information and to learn temporal sequences. TMAM uses a continuas representation and autoassociative feedback. A RBF model is used to implement the angle surface, which is liable for producing the angle information for the joint of the robot arm. The two surfaces compose the whole system which is liable for the trajectory planning and system control. Hence, the system receives information about the initial point and the target point, constructs the trajectory to reach the target point from the initial point and converts the spatial points which compose the trajectory, in values of joint angles. The obtained results show that TMAM model can recall, interpolate and extrapolate points in the sequences. The model has the ability of generating new trajectories and memorizing different size of sequences at the same time. This model also shows fast learning. The RBF model can recall the desired outputs with a small error and can receive a pattern which is formed by an unreachable final point and generate a set of angles which, in turn, represent a reachable final point.

The phenomenon of aggression has been of interest to psychologists for many years, and has resulted in a variety of theories which-attempt to explain its existence in man. Aggression is prevalent in our primary schools today and it was this observation which initiated the research project. The Relationship Theory was applied in order to gain insight into the life-world of the aggressive child and to explore possible causes which may originate from changes in our society. / Psychology of Education / M. Ed. with specialisation in Guidance and Counselling (Psychology of Education)

A contribuição e objetivo desta tese é desenvolver um modelo de redes neurais artificiais, baseado em princípios de memória associativa, capaz de resolver o problema de planejamento e controle ponto a ponto de trajetória de um braço mecânico imerso em um ambiente parcialmente conhecido e/ou sujeito a ruídos. O modelo proposto é formado por dois planos: plano seqüência temporal e plano ângulo. Para o plano seqüência temporal, o novo modelo proposto chamado de Memória Associativa Multidirecional Temporal (TMAM) é capaz de armazenar e recuperar n-tuplas de informações, lidar com informações ruidosas e/ou incompletas e aprender seqüências temporais. TMAM utiliza representação contínua e realimentação autoassociativa. O plano ângulo é formado pelo modelo RBF que é responsável por produzir as informações de ângulos das juntas do braço mecânico. A composição dos dois planos forma o sistema completo que é responsável pelo planejamento e controle ponto a ponto de trajetória. Em resumo, o sistema recebe informações do ponto origem e do ponto alvo, estabelece uma trajetória para atingir o ponto alvo a partir do ponto de origem e transforma os pontos espaciais da trajetória em valores de ângulos das juntas. Os resultados obtidos mostram que o modelo TMAM é capaz de recuperar, interpelar e extrapolar pontos nas seqüências, é capaz de gerar trajetórias, de memorizar seqüências de diferentes tamanhos e de lidar com duas trajetórias ao mesmo tempo. O modelo apresenta também rápido treinamento. O modelo RBF é capaz de recuperar as saídas desejadas apresentando um erro pequeno e é capaz de receber um padrão que apresenta um ponto final inatingível e gerar um conjunto de ângulos que representa um ponto final atingível. / The aim of this project is to develop an artificial neural networks model based on principles of associative memory. This neural network model must be able to solve the problem of trajectory planning and point to point control of a robot arm, which is located in a partially known and/or noisy environment. The proposed model is composed by two surfaces: the temporal sequence surface and the angle surface. For the temporal sequence surface the new propose model Temporal Multidirectional Associative Memmy (TMAM) is able to store and recall n-tuplas of information, to deal with noisy and/or incomplete information and to learn temporal sequences. TMAM uses a continuas representation and autoassociative feedback. A RBF model is used to implement the angle surface, which is liable for producing the angle information for the joint of the robot arm. The two surfaces compose the whole system which is liable for the trajectory planning and system control. Hence, the system receives information about the initial point and the target point, constructs the trajectory to reach the target point from the initial point and converts the spatial points which compose the trajectory, in values of joint angles. The obtained results show that TMAM model can recall, interpolate and extrapolate points in the sequences. The model has the ability of generating new trajectories and memorizing different size of sequences at the same time. This model also shows fast learning. The RBF model can recall the desired outputs with a small error and can receive a pattern which is formed by an unreachable final point and generate a set of angles which, in turn, represent a reachable final point.