In this paper we propose an architecture for a new task-level system, which we call TWAIN. Task-level programming attempts to simplify the robot programming process but requiring that the user specify only goals for the physical relationships among objects, rather than the motions needed to achieve those goals. A task-level specification is meant to be completely robot independent; no positions or paths that depend on the robot geometry or kinematics are specified by the user. We have two goals for this paper. Th is first is to present a more unified t reatment of some individual pieces of r esearch in task planning, whose r elationship has not previously been d escribed. The second is to provide a new framework for further research in task-planning. This is a slightly modified version of a paper that appeared in Proceedings of Soli d Modeling by Computers: from Theory to A pplications, Research laboratories Sympo sium Series, sponsored by General Motors, Warren, Michigan, September 1983.

The use of active compliance enables robots to carry out tasks in the presence of significant sensing and control errors. Compliant motions are quite difficult for humans to specify, however. Furthermore, robot programs are quite sensitive to details of geometry and to error characteristics and must, therefore, be constructed anew for each task. These factors motivate the need for automatic synthesis tools for robot programming, especially for compliant motion. This paper describes a formal approach to the synthesis of compliant motion strategies from geometric descriptions of assembly operations and explicit estimates of errors in sensing and control. A key aspect of the approach is that it provides correctness criteria for compliant motion strategies.

Le chapitre 1 est pour l'essentiel une brève introduction générale qui donne le contexte générale de la planification et présente l'organisation du document dans son ensemble et quelques uns des points clés retenus : robot humanoïde, environnement non statique, perception par vision artificielle, et représentation de cet environnement par grilles d'occupation. Dans le chapitre 2, après une revue de littérature bien menée, l'auteur propose de considérer les points de repère de l'environnement dès la phase de planification de chemin afin de rendre plus robuste l'exécution des déplacements en cas d'évolution de l'environnement entre le moment où la planification est menée et celui où le robot se déplace ( évolution étant entendu comme liée à une amélioration de la connaissance par mise à jour, ou due à un changement de l'environnement lui-même). Le concept est décrit et une formalisation proposée. Le chapitre 3 s'intéresse en détail à la planification dans le cas d'environnements dynamiques. Les méthodes existantes, nombreuses, sont tout d'abord analysées et bien présentées. Le choix est fait ici de décrire l'environnement comme étant décomposé en cellules, regroupant elles-mêmes des voxels, éléments atomiques de la représentation. L'environnement étant changeant, l'auteur propose de réévaluer le plan préétabli à partir d'une bonne détection de la zone qui a pu se trouver modifiée dans l'environnement. L'approche est validée expérimentalement en utilisant une des plateformes robotiques du LAAS qui dispose de bonnes capacités de localisation : le manipulateur mobile Jido étant à ce jour plus performant sur ce plan que l'humanoïde HRP2, c'est lui qui a été utilisé. Ces expérimentations donnent des indications concordantes sur l'efficacité de l'approche retenue. Notons également que la planification s'appuie sur une boite englobante de l'humanoïde, et non pas sur une représentation plus riche (multi-degré-deliberté). En revanche, c'est bien de planification pour l'humanoïde considéré dans toute sa complexité qu'il s'agit au chapitre 4 : on s'intéresse ici à tous les degrés de liberté du robot. L'auteur propose des évolutions de méthodes existantes et en particulier sur la manière de tirer profit de la redondance cinématique. L'approche est bien décrite et permet d'inclure une phase d'optimisation de la posture globale du robot. Des exemples illustrent le propos et sont l'occasion de comparaison avec d'autres méthodes. Le chapitre 5 s'intéresse à la manière de modéliser l'environnement, sachant qu'on s'intéresse ici au cas d'une perception par vision artificielle, et précisément au cas de l'humanoïde, robot d'assurer lui-même cette perception au fur et à mesure de son avancée dans l'environnement. On est donc dans le cadre de la recherche de la meilleure vue suivante qui doit permettre d'enrichir au mieux la connaissance qu'a le robot de son environnement. L'approche retenue fait à nouveau appel à la boite englobante de l'humanoïde et non à sa représentation complète ; il sera intéressant de voir dans le futur ce que pourrait apporter la prise en compte des degrés de liberté de la tête ou du torse à la résolution de ce problème. Le chapitre 6 décrit la phase d'intégration de tous ces travaux sur la plateforme HRP2 du LAAS-CNRS, partie importante de tout travail de roboticien. / This thesis starts by proposing a new framework for motion planning using stochastic maps, such as occupancy-grid maps. In autonomous robotics applications, the robot's map of the environment is typically constructed online, using techniques from SLAM. These methods can construct a dense map of the environment, or a sparse map that contains a set of identifiable landmarks. In this situation, path planning would be performed using the dense map, and the path would be executed in a sensor-based fashion, using feedback control to track the reference path based on sensor information regarding landmark position. Maximum-likelihood estimation techniques are used to model the sensing process as well as to estimate the most likely nominal path that will be followed by the robot during execution of the plan. The proposed approach is potentially a practical way to plan under the specific sorts of uncertainty confronted by a humanoid robot. The next chapter, presents methods for constructing free paths in dynamic environments. The chapter begins with a comprehensive review of past methods, ranging from modifying sampling-based methods for the dynamic obstacle problem, to methods that were specifically designed for this problem. The thesis proposes to adapt a method reported originally by Leven et al.. so that it can be used to plan paths for humanoid robots in dynamic environments. The basic idea of this method is to construct a mapping from voxels in a discretized representation of the workspace to vertices and arcs in a configuration space network built using sampling-based planning methods. When an obstacle intersects a voxel in the workspace, the corresponding nodes and arcs in the configuration space roadmap are marked as invalid. The part of the network that remains comprises the set of valid candidate paths. The specific approach described here extends previous work by imposing a two-level hierarchical structure on the representation of the workspace. The methods described in Chapters 2 and 3 essentially deal with low-dimensional problems (e.g., moving a bounding box). The reduction in dimensionality is essential, since the path planning problem confronted in these chapters is complicated by uncertainty and dynamic obstacles, respectively. Chapter 4 addresses the problem of planning the full motion of a humanoid robot (whole-body task planning). The approach presented here is essentially a four-step approach. First, multiple viable goal configurations are generated using a local task solver, and these are used in a classical path planning approach with one initial condition and multiple goals. This classical problem is solved using an RRT-based method. Once a path is found, optimization methods are applied to the goal posture. Finally, classic path optimization algorithms are applied to the solution path and posture optimization. The fifth chapter describes algorithms for building a representation of the environment using stereo vision as the sensing modality. Such algorithms are necessary components of the autonomous system proposed in the first chapter of the thesis. A simple occupancy-grid based method is proposed, in which each voxel in the grid is assigned a number indicating the probability that it is occupied. The representation is updated during execution based on values received from the sensing system. The sensor model used is a simple Gaussian observation model in which measured distance is assumed to be true distance plus additive Gaussian noise. Sequential Bayes updating is then used to incrementally update occupancy values as new measurements are received. Finally, chapter 6 provides some details about the overall system architecture, and in particular, about those components of the architecture that have been taken from existing software (and therefore, do not themselves represent contributions of the thesis). Several software systems are described, including GIK, WorldModelGrid3D, HppDynamicObstacle, and GenoM.

For over two decades, studies on task planning and its role in second language learners' oral performance have shown that the opportunity to plan for a task generally improves learners' speech (Ellis, 2005). It has been hypothesized that the opportunity to plan for a task reduces cognitive load during language processing, thus allowing learners to attend to various aspects of language, and that this enhanced attention, in turn, results in more successful task performance. However, one limitation to this task planning research to date it that most studies have examined the effects of planning before task performance, while largely ignoring the effects of planning that occur during task performance (Yuan & Ellis, 2003). Another limitation in planning research is that findings have been based exclusively on external observation and measurement of learners' oral production; we know little about what strategies learners use that may result in higher-quality speech. The participants in this study were intermediate and high-intermediate learners of Japanese. They were divided into experimental groups and performed a narrative task under different task conditions. Participants received a set of pictures and were asked to retell the story in Japanese. To examine the effects of planning on task performance, fluency, complexity, and accuracy in the participants' speech were analyzed. For the analysis of planning strategies, retrospective interviews were given to a group of participants from each planning group immediately after the task performance. The results revealed that there were no significant differences in participants' oral production across planning conditions, except in the area of lexical complexity (participants without a pre-task planning opportunity produced narrative stories with a greater variety of vocabulary than those who planned before the task). A trade-off effect between lexical complexity and accuracy was found when participants planned either before or during the task. Another trade-off effect was found between lexical complexity and fluency for the participants with on-line planning only. The analyses of strategy use showed that second language learners generally selected similar strategies regardless of planning conditions. These results provided important pedagogical implications and suggested useful future research directions.

京都大学 / 新制・課程博士 / 博士(情報学) / 甲第25425号 / 情博第863号 / 京都大学大学院情報学研究科知能情報学専攻 / (主査)教授 森 信介, 教授 中村 裕一, 教授 鹿島 久嗣 / 学位規則第4条第1項該当 / Doctor of Informatics / Kyoto University / DFAM

This work explores various methods teachers can use to promote high quality second language oral production. It consists of a review of empirical research and pedagogical implications related to the following factors: 1) Pre-task planning, 2) within-task planning, 3) task repetition, 4) task design, 5) formulaic sequences, 6) learner strategies, 7) form instruction, and 8) error correction. The work concludes with a consideration of issues within the literature and a brief summary of pedagogical implications. / text

This thesis presents an extension to the Temporal Fast Downward planning system that integrates motion planning in it and algorithms for generating two types of temporal macro operators expressible in PDDL2.1. The extension to the Temporal Fast Downward planning system includes, in addition to the integration of motion planning itself, an extension to the context-enhanced additive heuristic that uses information from the motion planning part to improve the heuristic estimate. The temporal macro operators expressible in PDDL2.1 are, to the author's knowledge, an area that is not studied within the context of plan repair before. Two types of temporal macro operators are presented along with algorithms for automatically constructing and using them when solving plan repair problems by replanning. Both the heuristic extension and the temporal macro operators were evaluated in the context of simulated unmanned aerial vehicles autonomously executing reconnaissance missions to identify targets and avoiding threats in unexplored areas. The heuristic extension was proved to be very helpful in the scenario. Unfortunately, the evaluation of the temporal macro operators indicated that the cost of introducing them is higher than the gain of using them for the scenario.

Over the past several years, much research has investigated the role of pre-task planning, including solitary, group, and teacher-led planning, on the variables of complexity, fluency, and accuracy in Second Language Acquisition (SLA) research. (Foster & Skehan, 1996; Gaillard, 2013; Geng & Ferguson, 2013). Additionally, other studies have investigated L2 learners' use of paragraphs and/or the role of conjunctions, i.e. transition words and expressions, in developing ideas and increasing cohesion (Mendelson, 2012; Rass, 2015). A gap remains, however, in seeing how pre-speaking and transition word activities together can promote proficiency in terms of text type, i.e. the move from word level speech and producing strings of sentences to paragraph level discourse. This study seeks to fill this gap by examining two teaching methods, namely Prelude to Conversation, or pre-speaking (Thompson, 2009), and transition word activities, to investigate the effect that these teaching methods have on increasing complexity and fluency among Intermediate-level learners of French. Complexity was measured by investigating the sub-components of total transition words, taught transition words, total clauses, words per clause, and total words. Fluency was measured by investigating the sub-components of time duration (total minutes) and words per minute. Furthermore, a case study illustrates the implications of increases in complexity and fluency for text type. Subjects were recruited from third semester French courses at Brigham Young University and were subsequently divided into three groups with each group receiving a different teaching method: Group 1 received transition word pre-activities, Group 2 received pre-speaking with a focus on content and forms needed to respond to the task, and Group 3 received a combination of both teaching methods. The study lasted four weeks with a Pre-Test in week one, followed by two weeks of treatments before completing the Post-Test in the fourth week. During the second and third weeks, each group received their respective treatments before responding to prompts that were identical for each group. Following the data collection, the speech samples were transcribed and analyzed for the sub-components of complexity and fluency. Results show, when comparing the Pre-Test to the Post-Test, that pre-speaking has a broader impact on complexity and fluency, either alone or when combined with transition word activities, impacting in particular total clauses, total words and response duration. When transition word activities were taught alone, there were greater gains in the use of taught transition words. The findings also demonstrate that even simply practicing providing oral responses regardless of treatment did help learners make overall increases that led to Post-Test responses (without scaffolding) that did not return to Pre-Test levels.

<p>It is often beneficial for an autonomous agent that operates in a complex environment to make use of different types of mathematical models to keep track of unobservable parts of the world or to perform prediction, planning and other types of reasoning. Since a model is always a simplification of something else, there always exists a tradeoff between the model’s accuracy and feasibility when it is used within a certain application due to the limited available computational resources. Currently, this tradeoff is to a large extent balanced by humans for model construction in general and for autonomous agents in particular. This thesis investigates different solutions where such agents are more responsible for balancing the tradeoff for models themselves in the context of interleaved task planning and plan execution. The necessary components for an autonomous agent that performs its abstractions and constructs planning models dynamically during task planning and execution are investigated and a method called DARE is developed that is a template for handling the possible situations that can occur such as the rise of unsuitable abstractions and need for dynamic construction of abstraction levels. Implementations of DARE are presented in two case studies where both a fully and partially observable stochastic domain are used, motivated by research with Unmanned Aircraft Systems. The case studies also demonstrate possible ways to perform dynamic abstraction and problem model construction in practice.</p> / Report code: LiU-Tek-Lic-2008:21.

Un intérêt croissant est aujourd'hui porté sur les robots capables de conduire des activités de collaboration d'une manière naturelle et efficace. Nous avons développé une architecture et un système qui traitent des aspects décisionnels de ce problème. Nous avons mis en oeuvre cette architecture pour traiter trois problèmes différents: le robot observateur, le robot équipier et enfin le robot instructeur. Dans cette thèse, nous discutons des défis et problématiques de la coopération homme-robot, puis nous décrivons l'architecture que nous avons développée et enfin détaillons sa mise oeuvre et les algorithmiques spécifiques à chacun des scénarios.Dans le cadre du scénario du robot observateur, le robot maintient un état du monde à jour au moyen d'un raisonnement géométrique effectué sur les données de perception, produisant ainsi une description symbolique de l'état du monde et des agents présents. Nous montrons également, sur la base d'un système de raisonnement intégrant des processus de décision de Markov (MDPs) et des réseaux Bayésiens, comment le robot est capable d'inférer les intentions et les actions futures de ses partenaires humain, à partir d'une observation de leurs mouvements relatifs aux objets de l'environnement. Nous identifions deux types de comportements proactifs : corriger les croyances de l'homme en lui fournissant l'information pertinente qui lui permettra de réaliser son but, aider physiquement la personne dans la réalisation de sa tâche, une fois celle-ci identifiée par le robot.Dans le cas du robot équipier, ce dernier doir réaliser une tâche en coopération avec un partenaire human. Nous introduisons un planificateur nommé Human-Aware Task Planner et détaillons la gestion par notre systeme du plan partagé par un composant appelé Plan Management component. Grâce à se système, le robot peut collaborer avec les hommes selon trois modalités différentes : robot leader, human leader, ou equal partners. Nous discutons des fonctions qui permettent au robot de suivre les actions de son partenaire humain et de vérifier qu'elles sont compatibles ou non avec le plan partagé et nous montrons comment le robot est capable de produire des comportements sûrs qui permettent de réaliser la tâche en prenant en compte de manière explicite la présence et les actions de l'homme ainsi que ses préférences. L'approche est fondée sur des processus décisionnels de Markov hiérarchisés avec observabilité mixte et permet d'estimer l'engagement de l'homme et de réagir en conséquence à différents niveaux d'abstraction. Enfin, nous discutions d'une approche prospective fondée sur un planificateur multi-agent probabiliste mettant en œuvre des MDPs et de sa pertinence quant à l'amélioration du composant de gestion de plan partagé.Dans le scénario du robot instructeur, nous détaillons les processus décisionnels qui permettent au robot d'adapter le plan partagé (shared plan) en fonction de l'état de connaissance et des désirs de son partenaire humain. Selon, le cas, le robot donne plus ou moins de détails sur le plan et adapte son comportement aux connaissances de l'homme ; Une étude utilisateur a également été menée permettant de valider la pertinence de cette approche.Finalement, nous présentons la mise en œuvre d'un robot guide autonome et détaillons les processu décisionnels que nous y avons intégrés pour lui permettre de guider des voyageurs dans un hall d'aéroport en s'adaptant au mieux au contexte et aux désirs des personnes guidées. Nous illustrons dans ce contexte des comportement adaptatifs et pro-actifs. Ce système a été effectivement intégré sur le robot Spencer qui a été déployé dans le terminal principal de l'aéroport d'Amsterdam (Schiphol). Le robot a fonctionné de manière robuste et satisfaisante. Une étude utilisateur a permis, dans ce cas également, de mesurer les performances et de valider le système. / There has been an increasing interest, in the last years, in robots that are able to cooperate with humans not only as simple tools, but as full agents, able to execute collaborative activities in a natural and efficient way. In this work, we have developed an architecture for Human-Robot Interaction able to execute joint activities with humans. We have applied this architecture to three different problems, that we called the robot observer, the robot coworker, and the robot teacher. After quickly giving an overview on the main aspects of human-robot cooperation and on the architecture of our system, we detail these problems.In the observer problem the robot monitors the environment, analyzing perceptual data through geometrical reasoning to produce symbolic information.We show how the system is able to infer humans' actions and intentions by linking physical observations, obtained by reasoning on humans' motions and their relationships with the environment, with planning and humans' mental beliefs, through a framework based on Markov Decision Processes and Bayesian Networks. We show, in a user study, that this model approaches the capacity of humans to infer intentions. We also discuss on the possible reactions that the robot can execute after inferring a human's intention. We identify two possible proactive behaviors: correcting the human's belief, by giving information to help him to correctly accomplish his goal, and physically helping him to accomplish the goal.In the coworker problem the robot has to execute a cooperative task with a human. In this part we introduce the Human-Aware Task Planner, used in different experiments, and detail our plan management component. The robot is able to cooperate with humans in three different modalities: robot leader, human leader, and equal partners. We introduce the problem of task monitoring, where the robot observes human activities to understand if they are still following the shared plan. After that, we describe how our robot is able to execute actions in a safe and robust way, taking humans into account. We present a framework used to achieve joint actions, by continuously estimating the robot's partner activities and reacting accordingly. This framework uses hierarchical Mixed Observability Markov Decision Processes, which allow us to estimate variables, such as the human's commitment to the task, and to react accordingly, splitting the decision process in different levels. We present an example of Collaborative Planner, for the handover problem, and then a set of laboratory experiments for a robot coworker scenario. Additionally, we introduce a novel multi-agent probabilistic planner, based on Markov Decision Processes, and discuss how we could use it to enhance our plan management component.In the robot teacher problem we explain how we can adapt the plan explanation and monitoring of the system to the knowledge of users on the task to perform. Using this idea, the robot will explain in less details tasks that the user has already performed several times, going more in-depth on new tasks. We show, in a user study, that this adaptive behavior is perceived by users better than a system without this capacity.Finally, we present a case study for a human-aware robot guide. This robot is able to guide users with adaptive and proactive behaviors, changing the speed to adapt to their needs, proposing a new pace to better suit the task's objectives, and directly engaging users to propose help. This system was integrated with other components to deploy a robot in the Schiphol Airport of Amsterdam, to guide groups of passengers to their flight gates. We performed user studies both in a laboratory and in the airport, demonstrating the robot's capacities and showing that it is appreciated by users.