jeudi 18 mars 2010

Ambient intelligence:




    Ambient intelligence is a term equating evolution of computing, from computers to everyday objects. Also referred to this change in the terms of ubiquitous computing, reflecting its ubiquity in the environment. Waldner shows the evolution of computing to ambient intelligence in four key steps:
-The Birth of the computer (1960).
-The Birth of computer communication networks (such as the Internet 1970)
-The Mobile computing devices by managing portable energy sources (mobile systems 1990/2000);
-The Proliferation of computing devices in the environment (ubiquitous systems) until their density is large enough to develop skilled functions of "smart" (ambient intelligence 2000/2010).

Time evolution of computing, from birth to ambient computing

    Ambient intelligence is the result of the proposal in 1991 by Mark Weiser. In its work, the author explains that computers tend to miniaturize and to spread, illustrating the passage of the "frames hand" (a computer shared by multiple people) to personal computers (one computer per person). He enters and sees what could be the computer of the future, for miniaturization and networking of computer devices. Weiser illustrates this concept by the "tabs", then developing in his laboratory, which are the equivalent of digital tablets industrialized today.

    In his vision, becoming more and more numerous and miniaturized computers will melt in our daily lives, whether physically or through use, to become invisible. Especially for developing this final step in order to then formalize the concept of smart home.

    The increase in the density of its IT resources corollary to increase the information density on the environment. Ambient intelligence is the consideration of environmental data (perception) in ubiquitous computing. Environmental data to determine the context in which the system operates.


   Atmosphere; Ambient: material or moral atmosphere that surrounds a person, a meeting
persons; around all sides, is the place where one is [Webster's].
Intelligence: (a) faculty perceive, reason and act on its environment in order to achieve or maintain a state (or objective). (B) Faculty interact with its environment in order to achieve or maintain a state (or objective).


Pyramid context proposed by Coutaz, Crowley and Rey

    This pyramidal structure shows that ambient intelligence is not only due to the instrumentation of the environment with a number of sensors. The exploitation of these sensor data goes through stages, involving different areas of research, we propose to detail below.

    Ambient intelligence involves the concept of environment. It is indeed to provide a given environment of digital capabilities (urban space, office, commercial spaces.): Perception abilities (from various sensors), processing capacity and response capabilities (by various effectors). Ambient intelligence thus involves many areas, especially in the field of sensor networks (and implicitly the actuator networks), the field of human-machine interaction and the field of artificial intelligence

    Networking and computer miniaturization are developments that support ambient intelligence. However, these developments imply a significant increase in the number of points using the network infrastructure (projections converge to tens of billions of objects connected by 2020). Networking systems increasingly miniaturized raises several issues such as mobility, energy dependence and the evolution of communications infrastructure. Consequently, a theme is dedicated to the telecommunication protocols satisfying these issues. Thus, telecommunications and network specialists have made ad hoc networks without son (mobile) and low energy consumption of a major research topic today leading to new network technologies such as ZigBee and 6LoWPAN.

Evolution of the number of connected objects 2020 Source: CiscoIBSG

    This theme is also illustrated with a field of research called "Internet of Things". In this paradigm, each object is identified, connected and integrated into a system, thereby forming a data network. This data network creates an intelligent environment. To do this, near field technology can identify objects (RFID, Bluetooth, QRCodes, NFC.), Objects can also communicate (6LoWPAN). These objects are then integrated into a middleware (or middleware), allowing to abstract the data and provide application interfaces. One such example is illustrated by middleware OSGi community.

    The field of human-computer interaction (HCI) is to explore ways of interaction between technologies developed and humans. It tends to a natural vision of Human Computer Interaction in opposition to type interfaces "WIMP". This goal is illustrated by the development of gestural interfaces, tactile and tangible (multimodal) that blur traditional Fulani little keyboards and mice. Ambient intelligence provides additional elements to improve usability criteria for these interfaces, such as the generation of contextual interfaces. These interfaces can be adapted to the context in which they are subject. For example, a weather service interface presents the user with the weather of where it is located, by locating the automatic and transparent way.

    An example of man-machine interaction, quiet computing, is the "LiveWire" developed by Xerox Jeremijenko, reported in (Ishiietal., 1997; Reignier, 2010). This is a step motor, connected to a string which, at every transit network packet emits an audible and visual information. This engine makes it possible to spread the network traffic condition in the environment close to the user. It detects more naturally any abnormal behavior of the motor, and therefore the network.



    Another example is the tangible interfaces, where the user manipulates physical objects from daily life in order to establish a dialogue with a machine. Ishii (Ishiietal. 1997) illustrated this concept with several examples, including "metaDESK" and "trans BOARD", a desk and a whiteboard that constitute both an interaction surface. The user can then use these surfaces, using physical objects: lenses or geographical forms for the "metaDesk" or so felt for the "metaBOARD". In the case of "metaDesk" information are redirected graphically on a lens equipped with an LCD screen, depending on the position or while the movement of objects (displacement, rotation, etc.). In the case of "metaBOARD" table detects contact with the felt which then makes it capable of storing forms and texts in digital form, dematerializing and information.

     Artificial intelligence (AI) is a field that aims to reproduce human intelligence in computers. This area is known to the public because it feeds regularly works of science fiction, such as the films "2001: A Space Odyssey" by Stanley Kubrick (1968), "WarGames" John Badhamen 1983 or more recently "Artificial IAIntelligence" Steven Spielberg (2001). John McCarthy and Marvin Minsky are considered the fathers of the discipline that formed shortly after the birth of computing.
Work from this area have been major advances, which are now increasingly used. For example, many algorithms use machine learning, in various forms. Thus, a machine can be adapted depending on learned data during its use, as shown by the approaches based on the statistical learning and probabilistic reasoning.

    The work in artificial intelligence have been particularly developed in the field of robotics. One of the best known implementation is the Bayesian programming mobile robots examples. The problem of Lebeltel is, in this example, to direct a robot (motors driving the wheels controls) with a limited number of sensors (light sensor). The complexity of the problem is the uncertainty of the environment (soil type, obstacles), sensors and effectors (wheel spin). He then proposed a probabilistic approach, where sensors are associated probability distributions, the melting quantifies actions to effectors. These distributions can be specified by an expert.

    Ambient intelligence and automation have developed in parallel and converge toward the smart home. "I smart housing can be considered as an application area of ​​ambient intelligence, it is difficult to place it in relation to home automation. The above definitions are emerging the following: home automation is "desired". It is not essential to the inhabitant, testifies in his failures and imaginary descriptions but gives extra comfort by technological habitat management. Ambient intelligence is, in turn, more "experienced", it is the result of an almost natural evolution of technology, directed evolution by the miniaturization of electronics and is introducing patchy in the habitat.


Aucun commentaire:

Enregistrer un commentaire