Transcript 슬라이드 1
Ambient Intelligence: A New Multidisciplinary Paradigm Written by Paolo Remagnino, Gian Luca Foresti IEEE TRANSACTIONS ON SYSTEMS, MAN, AND CYBERNETICS PART A: SYSTEMS AND HUMANS, VOL. 35, NO. 1, JANUARY 2005 Presented by Dongjoo Lee IDS Lab., CSE, SNU Ambient Intelligence (AmI) New paradigm that supports the design of the next generation of intelligent systems and introduces novel means of communication between human, machine, and the surrounding environment. Computers disappear in the background while users moves into the foreground in complete control of the augmented environment Copyright 2008 by CEBT AmI System Assist the user by autonomously interpreting their intentions Enhancing the training of professional skills Making simpler and more pleasant the life Based on Modularity Low-power devices Distributed and high bandwidth heterogeneous networks of sensors and actuators Rely on machine learning research Complex models is unrealistic Mapping of sensory information onto behaviors is too complex Copyright 2008 by CEBT Multidisciplinary distributed intelligence software design ethics data and information communication law computer wearables computer vision speech recognition information fusion social sciences hardware design robotics Copyright 2008 by CEBT What would be able to realize AmI? Distributed Intelligence Agents Pervasive and distributed layer of intelligence Hardware Design Hardware technologies have to be considered to enhance people’s lives in smart spaces Information Understanding Machine understanding Computer vision, speech recognition, and so on Sensor modeling, deployment and combining of distributed sensor information Communication Modeling Intelligent layer have to be built on top of a robust seamless communication infrastructure Copyright 2008 by CEBT AmI Environments Sensors and Devices User Seamless Communication Infrastructure Intelligence Doctoral Thesis, Dongjoo Lee Copyright 2008 by CEBT 6 Special Issue on AmI Detection of the users within AmI spaces, in particular through video cameras and applying computer vision techniques Architectures for AmI Human factors research for user-requirements design and quality assessment Sensors and communication required for the infrastructure of an AmI system Computer vision research, linked with one or more camera sensors, including data fusion problems Artificial intelligence solutions, including scene understanding and creation of simulation AmI solutions for different application domains including health, practical skills training, public spaces management, etc Performance evaluation: can measures be defined to evaluate and compare AmI systems Copyright 2008 by CEBT 1) Affective State Detection with DBN Active Affective State Detection and User Assistance of With Dynamic Bayesian Networks, Xiangyang Li,and Qiang Ji Dynamic Bayesian Network Dynamically model and recognize affective states of the user Provide the appropriate assistance Copyright 2008 by CEBT 2) Multi Agent Intelligent Building Control and Learning of Ambience by an Intelligent Building, Ueli Rutishauser, Josef Joller, and Rodney Douglas Intelligent Buildings Dynamic reconfiguration of space and function Sensors and effectors are linked by means of fuzzy rules and the agents communicate with one another by asynchronous messaging Unsupervised online real-time learning algorithm Fuzzy rule derived from very sparse data in a nonstationary environment Copyright 2008 by CEBT 3) Video Security Video Security for Ambient Intelligence, Lauro Snidaro, Christian Micheloni, and Christian Chiavedale Security of people in intelligent building Tracking and counting people through multiple video sensors – Preventing entry to dangerous or nonauthorized areas – Regulate subsystems Heating Ventilation Air conditioning Main Processing Step Copyright 2008 by CEBT 4) Visual-based Surveillance System Prismatica: Towards Ambient Intelligence in Public Transport Environments, Sergio A. Velastin, Boghos A. Boghossian, Benny Ping Lai Lo, Jie Sun, and Maria Alicia Vicencio-Silva System architecture considers distributed nature of the detection process The system components have been implemented, integrated and tested in a real metropolitan railway environment Abnormal direction of motion Intrusion near the edge of a platform Copyright 2008 by CEBT 5) Planning Mechanisms for Goal-oriented Behavior What Planner for Ambient Intelligence Applications?, Francesco Amigoni, Nicola Gatti, Carlo Pinciroli, and Manuel Roveri Distributed hierarchical task network approach Characterized the planning problem within AmI Combine centralized and distributed features and the ability of the proposed planner to adapt the planning process and its results to the capabilities of the devices currently connected to the system Diabetic patient scenario Complete task network for the goal CheckAndRequest (Insulin) Copyright 2008 by CEBT 6) An Interactive Space An Interactive Space That Learns to Influence Human Behavior, Kynan Eng, Rodney J. Douglas, and Paul F. M. J. Verschure Distributed Adaptive Control Based on the animal learning paradigms Applied to the learning of effective cues for guiding visitors in a given direction Copyright 2008 by CEBT 7) Structured Context Analysis Structured Context Analysis Techniques in Biologically Inspired Ambient Intelligent Systems, Luca Marchesotti, Stefano Piva, and Carlo Regazzoni Neurobiologic Model for AmI A complex event classification is obtained through the fusion of heterogeneous data coming from a set of sensors Self-organizing map (SOM) Copyright 2008 by CEBT 8) Probabilistic Posture Classification Probabilistic Posture Classification for Human Behavior Analysis, Rita Cucchiara, Costantino Grana, Andrea Prati, and Roberto Vezzani Automated system for human behavior analysis Classify the posture of a person and detecting corresponding events and alarm situations like a fall 95% accuracy!!. Copyright 2008 by CEBT 9) Figuring Out State of Intelligent Spaces Dynamic Context Capture and Distributed Video Arrays for Intelligent Spaces, Mohan Manubhai Trivedi, Kohsia Samuel Huang, and Ivana Mikic Context of the Indoor Intruder detection Multiple person tracking Body pose and posture analysis Person identification Human body modeling and movement analysis Copyright 2008 by CEBT 10) Fuzzy Embedded Approach for AmI A Fuzzy Embedded Agent Based Approach for Realizing Ambient Intelligence in Intelligent Inhabited Environments, Faiyaz Doctor, Hani Hagras, and Victor Callaghan Unsupervised, data-driven, fuzzy technique that is used for extracting fuzzy membership functions and rules that represent particular user’s behaviors in the environments Used interface devices PC, iPAQ, mobile phone, iFridge Copyright 2008 by CEBT 11) Mobile Sensor Deployment Energy-Efficient Deployment of Intelligent Mobile Sensor Networks, Nojeong Heo and Pramod K. Varshney Copyright 2008 by CEBT 12) Critical Situation Identification Health-Status Monitoring Through Analysis of Behavioral Patterns, Tracy Barger, Donald Brown, and Majd Alwan Virtual maid for tracking movements of the elderly and alerting when a critical situation is identified. Assume deployment of a number of motion sensors Copyright 2008 by CEBT Discussion AmI Various sensors in a distributed way Integrate heterogeneous information to interpret the current situation and do something useful to user. Machine learning approaches, that are unsupervised and data driven Uncertain and fuzzy notion Layered and structured context interpretation It’s really multidisciplinary. My discussion Still we have to try to know what a user want in a certain situation and define minimal level of abstraction to implement it in our daily life. Let’s realize one in a way we are familiar with and we can do well. Copyright 2008 by CEBT