Submitted by Fabio Sobreira Campos da Costa (fabio.sobreira@ufpe.br) on 2017-07-12T13:16:29Z No. of bitstreams: 2 license_rdf: 811 bytes, checksum: e39d27027a6cc9cb039ad269a5db8e34 (MD5) tese_doutorado_amilcar-07-2016_versao-cd (1).pdf: 2101060 bytes, checksum: 21d268c59ad60238bce0cde073e6f3cd (MD5) / Made available in DSpace on 2017-07-12T13:16:29Z (GMT). No. of bitstreams: 2 license_rdf: 811 bytes, checksum: e39d27027a6cc9cb039ad269a5db8e34 (MD5) tese_doutorado_amilcar-07-2016_versao-cd (1).pdf: 2101060 bytes, checksum: 21d268c59ad60238bce0cde073e6f3cd (MD5) Previous issue date: 2016-03-10 / A popularização de tecnologias de captura de dados geolocalizados aumentou a quantidade de dados de trajetórias disponível para análise. Trajetórias de objetos móveis são geradas a partir das posições de um objeto que se move durante um certo intervalo de tempo no espaço geográfico. Para diversas aplicações é necessário que as trajetórias sejam divididas em partições menores, denominadas segmentos, que representam algum comportamento relevante para a aplicação. A literatura reporta diversos trabalhos que propõem a segmentação de trajetórias. Entretanto, pouco se discute a respeito de quais algoritmos são mais adequados para um domínio ou quais valores de parâmetros de entrada fazem com que um algoritmo obtenha o melhor desempenho neste mesmo domínio. A grande maioria dos algoritmos de segmentação de trajetórias utiliza critérios pré-definidos para realizar esta tarefa. Poucos trabalhos procuram utilizar critérios nos quais não se sabe a priori que tipos de segmentos são gerados, sendo esta questão pouco explorada na literatura. Outra questão em aberto é o uso de exemplos para induzir um algoritmo de segmentação a encontrar segmentos semelhantes a estes exemplos em outras trajetórias. Esta proposta de tese objetiva resolver estas questões. Primeiro, são propostos os métodos GEnetic Algorithm based on Roc analysis (GEAR) e o Iterated F-Race for Trajectory Segmentation Algorithms (I/F-Race-TSA), que são métodos para auxiliar na escolha da melhor configuração (i.e. valores de parâmetros de entrada) de algoritmos de segmentação de trajetórias. Segundo, é proposto o Greedy Randomized Adaptive Search Procedure for Unsupervised Trajectory Segmentation (GRASP-UTS), com o objetivo de resolver o problema de segmentação de trajetórias quando o critério de segmentação não é previamente definido. Por último, propomos o GRASP for Semi-supervised Trajectory Segmentation (GRASP-SemTS). O GRASP-SemTS usa exemplos para induzir a tarefa de segmentação a encontrar segmentos semelhantes em outras trajetórias. Foram conduzidos experimentos com os métodos e algoritmos propostos para domínios distintos e para trajetórias reais de objetos móveis. Os resultados mostraram que ambos os métodos GEAR e I/F-Race-TSA foram capazes de calibrar automaticamente os parâmetros de entrada de algoritmos de segmentação de trajetórias para um dado domínio de aplicação. Os algoritmos GRASP-UTS e GRASP-SemTS obtiveram melhor desempenho quando comparados a outros algoritmos de segmentação de trajetórias da literatura contribuindo assim com importantes resultados para a área. / The popularization of technologies for geolocated data increased the amount of trajectory data available for analysis. Moving objects’ trajectories are generated from the positions of an object that moves in the geographical space during a certain amount of time. For many applications, it is necessary to partition trajectories into smaller pieces, named segments, which represent a relevant behavior to the application point of view. The literature reports many studies that propose trajectory segmentation approaches. However, there is a lack of discussions about which algorithm is more likely to be applied in a domain or which values of its input parameters obtain the best performance in the domain. Most algorithms for trajectory segmentation use pre-defined criteria to perform this task. Only few works make use of criteria where the characteristics of the segment are not known a priori and this topic is not well explored in the literature. Another open question is how to use a small amount of labeled segments to induce a segmentation algorithm in order to find such kind of behaviors into unseen trajectories. This thesis proposal aims to solve these questions. First, we propose the GEnetic Algorithm based on Roc analysis (GEAR) and the Iterated F-Race for Trajectory Segmentation Algorithms (I/F-RaceTSA), which are methods that are able to find the best configuration (i.e. input parameter values) of algorithms for trajectory segmentation. Second, we propose a Greedy Randomized Adaptive Search Procedure for Unsupervised Trajectory Segmentation (GRASP-UTS) aiming to solve the trajectory segmentation problem when the criteria is not determined a priori. Last, we propose the GRASP for Semi-supervised Trajectory Segmentation (RGRASP-SemTS). The GRASP-SemTS solves the problem of using a small amount of labeled data to induce the trajectory segmentation algorithm to find such behaviors into unseen trajectories. Experiments were conducted with the methods and algorithms algorithms using real world trajectory data. Results showed that GEAR and I/F-Race-TSA are capable of finding automatically the input parameter values for a domain. The GRASP-UTS and GRASP-SemTS obtained a better performance when compared to other segmentation algorithms from literature, contributing with important results for this field.

The increase in population, accompanied by an increase in the availability of travel opportunities have kindled the interest in understanding how people make use of the space around them and their opportunities. Understanding the travel behaviour of individuals and groups is difficult because of two main factors: the travel behaviour's wide coverage, which encompasses different research areas, all of which model different aspects of travel behaviour, and the difficulty of obtaining travel diaries from large groups of respondents, which is imperative for analysing travel behaviour and patterns. A travel diary allows an individual to describe how she performed her activities by specifying the destinations, purposes and travel modes occurring during a predefined period of time. Travel diaries are usually collected during a large-scale survey, but recent developments show that travel diaries have important drawbacks such as the collection bias and the decreasing response rate. This led to a surge of studies that try to complement or replace the traditional declaration-based travel diary collection with methods that extract travel diary specific information from trajectories and auxiliary datasets. With the automation of travel diary generation in sight, this thesis presents a suitable method for collecting data for travel diary automation (Paper I), a framework to compare multiple travel diary collection systems (Paper II), a set of relevant metrics for measuring the performance of travel mode segmentation methods (Paper III), and applies these concepts during different case studies (Paper IV). / <p>QC 20160525</p>

Current methods of robot task teaching and learning have several limitations: highly-trained personnel are usually required to teach robots specific tasks; service-robot systems are limited in learning different types of tasks utilizing the same system; and the teacher’s expertise in the task is not well exploited. A human-inspired robot-task teaching and learning method is developed in this research with the aim of allowing general users to teach different object-manipulation tasks to a service robot, which will be able to adapt its learned tasks to new task setups. The proposed method was developed to be interactive and intuitive to the user. In a closed loop with the robot, the user can intuitively teach the tasks, track the learning states of the robot, direct the robot attention to perceive task-related key state changes, and give timely feedback when the robot is practicing the task, while the robot can reveal its learning progress and refine its knowledge based on the user’s feedback. The human-inspired method consists of six teaching and learning stages: 1) checking and teaching the needed background knowledge of the robot; 2) introduction of the overall task to be taught to the robot: the hierarchical task structure, and the involved objects and robot hand actions; 3) teaching the task step by step, and directing the robot to perceive important state changes; 4) demonstration of the task in whole, and offering vocal subtask-segmentation cues in subtask transitions; 5) robot learning of the taught task using a flexible vote-based algorithm to segment the demonstrated task trajectories, a probabilistic optimization process to assign obtained task trajectory episodes (segments) to the introduced subtasks, and generalization of the taught task trajectories in different reference frames; and 6) robot practicing of the learned task and refinement of its task knowledge according to the teacher’s timely feedback, where the adaptation of the learned task to new task setups is achieved by blending the task trajectories generated from pertinent frames. An agent-based architecture was designed and developed to implement this robot-task teaching and learning method. This system has an interactive human-robot teaching interface subsystem, which is composed of: a) a three-camera stereo vision system to track user hand motion; b) a stereo-camera vision system mounted on the robot end-effector to allow the robot to explore its workspace and identify objects of interest; and c) a speech recognition and text-to-speech system, utilized for the main human-robot interaction. A user study involving ten human subjects was performed using two tasks to evaluate the system based on time spent by the subjects on each teaching stage, efficiency measures of the robot’s understanding of users’ vocal requests, responses, and feedback, and their subjective evaluations. Another set of experiments was done to analyze the ability of the robot to adapt its previously learned tasks to new task setups using measures such as object, target and robot starting-point poses; alignments of objects on targets; and actual robot grasp and release poses relative to the related objects and targets. The results indicate that the system enabled the subjects to naturally and effectively teach the tasks to the robot and give timely feedback on the robot’s practice performance. The robot was able to learn the tasks as expected and adapt its learned tasks to new task setups. The robot properly refined its task knowledge based on the teacher’s feedback and successfully applied the refined task knowledge in subsequent task practices. The robot was able to adapt its learned tasks to new task setups that were considerably different from those in the demonstration. The alignments of objects on the target were quite close to those taught, and the executed grasping and releasing poses of the robot relative to objects and targets were almost identical to the taught poses. The robot-task learning ability was affected by limitations of the vision-based human-robot teleoperation interface used in hand-to-hand teaching and the robot’s capacity to sense its workspace. Future work will investigate robot learning of a variety of different tasks and the use of more robot in-built primitive skills.

Current methods of robot task teaching and learning have several limitations: highly-trained personnel are usually required to teach robots specific tasks; service-robot systems are limited in learning different types of tasks utilizing the same system; and the teacher’s expertise in the task is not well exploited. A human-inspired robot-task teaching and learning method is developed in this research with the aim of allowing general users to teach different object-manipulation tasks to a service robot, which will be able to adapt its learned tasks to new task setups. The proposed method was developed to be interactive and intuitive to the user. In a closed loop with the robot, the user can intuitively teach the tasks, track the learning states of the robot, direct the robot attention to perceive task-related key state changes, and give timely feedback when the robot is practicing the task, while the robot can reveal its learning progress and refine its knowledge based on the user’s feedback. The human-inspired method consists of six teaching and learning stages: 1) checking and teaching the needed background knowledge of the robot; 2) introduction of the overall task to be taught to the robot: the hierarchical task structure, and the involved objects and robot hand actions; 3) teaching the task step by step, and directing the robot to perceive important state changes; 4) demonstration of the task in whole, and offering vocal subtask-segmentation cues in subtask transitions; 5) robot learning of the taught task using a flexible vote-based algorithm to segment the demonstrated task trajectories, a probabilistic optimization process to assign obtained task trajectory episodes (segments) to the introduced subtasks, and generalization of the taught task trajectories in different reference frames; and 6) robot practicing of the learned task and refinement of its task knowledge according to the teacher’s timely feedback, where the adaptation of the learned task to new task setups is achieved by blending the task trajectories generated from pertinent frames. An agent-based architecture was designed and developed to implement this robot-task teaching and learning method. This system has an interactive human-robot teaching interface subsystem, which is composed of: a) a three-camera stereo vision system to track user hand motion; b) a stereo-camera vision system mounted on the robot end-effector to allow the robot to explore its workspace and identify objects of interest; and c) a speech recognition and text-to-speech system, utilized for the main human-robot interaction. A user study involving ten human subjects was performed using two tasks to evaluate the system based on time spent by the subjects on each teaching stage, efficiency measures of the robot’s understanding of users’ vocal requests, responses, and feedback, and their subjective evaluations. Another set of experiments was done to analyze the ability of the robot to adapt its previously learned tasks to new task setups using measures such as object, target and robot starting-point poses; alignments of objects on targets; and actual robot grasp and release poses relative to the related objects and targets. The results indicate that the system enabled the subjects to naturally and effectively teach the tasks to the robot and give timely feedback on the robot’s practice performance. The robot was able to learn the tasks as expected and adapt its learned tasks to new task setups. The robot properly refined its task knowledge based on the teacher’s feedback and successfully applied the refined task knowledge in subsequent task practices. The robot was able to adapt its learned tasks to new task setups that were considerably different from those in the demonstration. The alignments of objects on the target were quite close to those taught, and the executed grasping and releasing poses of the robot relative to objects and targets were almost identical to the taught poses. The robot-task learning ability was affected by limitations of the vision-based human-robot teleoperation interface used in hand-to-hand teaching and the robot’s capacity to sense its workspace. Future work will investigate robot learning of a variety of different tasks and the use of more robot in-built primitive skills.