Skip to Main content Skip to Navigation
Theses

Assistance robotisée à la personne en environnement coopérant

Abstract : The meeting of ambient intelligence and robotics support gave resulting in the so-called ambient robotic assistance, designed to assist a person with reduced autonomy. It relies on the existence of communicating objects networks in the environment of the person to decline a set of services and teleservices to facilitate the daily life of that person and his entourage. One or more robots may be present in this environment. A recent scientific community was built around the ubiquitous robotics. While previous generations of robots are designed to perform specific tasks and built as an independent unit, the new generation is ubiquitous. The autonomy of the robot is, in this context, obtained by a close interaction between the robot and communicating ambient environment. Until recent years the robot progress in a hostile environment that did not facilitate the task. In the context of ambient intelligence, communicating objects in the environment can play a "facilitator" in helping the robot to locate, to browse and to search object, and so on. Conversely, the robot can be seen as a communicating object that is put to use by other services than person with reduced autonomy assistance. The first objective of this thesis is to propose computer architecture for the robot and communicating objects cooperation, in the environment such as sensors or actuators. One difficulty is that the robot is moving in a saturated environment, which changes so unpredictably in the sense that, for example, ambient sensors may be available or unavailable, accessible or inaccessible. It is therefore necessary to have mechanisms to adapt to different or unexpected situations to permit the robot succeeded in his mission, to move from point A to point B, where possible, taking account constraints such as degree of urgency or level of accuracy. We also considered another ethical constraint in the robotics projects home assistance, which is the degree of intrusion. The intrusion of ambient intelligence capable of acting and perceiving, in the presence of the person, causes discomfort for it and his entourage. As part of this thesis, we evaluate our computer architecture for the robot and communicating objects cooperation, as well as coping mechanisms with robot localization. The aim is to provide localization result, even if it is either degraded and this, regardless of the situation. To meet this result, we introduced the concept of effect. The system seeks to approximate as closely as possible the requested effect. This is the acceptable solution within the meaning of certain criteria, which is used. This is not necessarily the optimal solution. The contributions of this first part are as follows. We propose robot-environment cooperative architecture consists of a multi-agent system (MAS) based on a knowledge base and a gateway that interacts with the ambient environment including communicating objects which is the robot and with a user interface. Our contributions concern MAS structure and of the knowledge base based on ontology's formalism. To address the problem of adaptation to the context on the one hand, and respect the level of discomfort of the user (intrusiveness MAS) on the other hand, we propose a solution based on training coalitions of agents AA-CRE (ambient agents for robot-environment cooperation) according to the desired effect. The second goal of this thesis is to propose robot-environment communicator tracking method. Our objective is to locate with certainty the robot in an area of habitat. The dimensions of this zone can vary widely. These dimensions depend on the needs of the task, even tasks sequence that the robot expects to achieve at this time, but also stem from factors such as the dispersal of sensors in the environment, quality of available measures. We assume that with a good knowledge of the situation, the certainty of being in this area with a known accuracy, will allow the robot to adopt the best strategy to ensure its mission which is to find the person in its habitat. Also, the position and orientation of the robot must be given with guaranteed accuracy. It is less accurate than the certainty that the robot belongs to the area that matters. Although the approach of localization is based on the ambition to use the measurements from a sensor network, the most varied possible in the environment or on the robot, it is not intended, at least in its basic answer to all situations. We will specify the scope of its use. One of the advantages of this method is to be able to use measurements from sensors of the robot and the environment. The measures are known angles with some uncertainty. The localization is obtained by multilateration. The contributions of this second part are as follows. We propose a localization method based on the principle of multiangulation. Solving the problem is obtained by set theoretic approach, specifically the interval analysis. As the size of the state vector is less than or equal to three, the calculations are relatively simple and quick (the simulation is performed in Matlab, a realistic assessment should be performed to estimate the execution time). The simple model chosen for the sensors makes it possible to integrate a wide variety of sensors, the crudest to the most complex, if it delivers a corner measurement. The version of the algorithm proposed in this thesis aims on the one hand, to show the interest of the intervals approach, and secondly, to have a localization method which will be used by sensor agents of SMA described above. Improvements can be made on several points. First, the outliers that violate the assumption of error bounded are not addressed. There are tracks in the literature about it. Moreover, it was considered, to simplify the presentation of the approach, that the localization of sensors and markers of the environment was known. But it is easy to introduce uncertainty in the algorithm.
Document type :
Theses
Complete list of metadatas

Cited literature [76 references]  Display  Hide  Download

https://tel.archives-ouvertes.fr/tel-00674653
Contributor : Frédéric Davesne <>
Submitted on : Monday, February 27, 2012 - 6:26:33 PM
Last modification on : Tuesday, June 30, 2020 - 11:56:08 AM
Long-term archiving on: : Monday, May 28, 2012 - 3:00:47 AM

Identifiers

  • HAL Id : tel-00674653, version 1

Collections

Citation

Antonio Andriatrimoson. Assistance robotisée à la personne en environnement coopérant. Automatique / Robotique. Université d'Evry-Val d'Essonne, 2012. Français. ⟨tel-00674653⟩

Share

Metrics

Record views

522

Files downloads

1475