neuron.csie.ntust.edu.twhomeworkNSCEUdataH.M.Le

Document Sample
neuron.csie.ntust.edu.twhomeworkNSCEUdataH.M.Le Powered By Docstoc
					TU/e: Technische Universiteit
         Eindhoven
   智慧結構、材料與空間生活 IN
      愛因霍分科技大學
                OUTLINE
• 愛因霍芬科技大學簡介
• DDSS Research programme
  – MAS in Collabortive Design
  – Human behaviour simulation
  – Measuring Housing Preferences Using Virtual
    Reality and Bayesian Belief Networks
  – 4D CAD
       愛因霍芬科技大學簡介
•   成立於1956年(今年50歲了)
•   荷蘭的三所工科大學之一(荷蘭大學都是國立的)
•   1998年被德國評定為歐洲最好的工科大學
•   世界排名301~400大學 (根據上海交大排名)等同於清大
•   八個院系(建築系、電子工程、化學及化學工程、工業工程及管理科學、
    應用物理、機械工程、哲學及社會科學、數學及電腦科學)

• 大學部約6000人,碩士生約200人,博士生約450
  人;3000多名教職員工,300多位教授
  DDSS Research programme
• In Eindhoven University of Technology, DDSS is
  the name for several of our activities.
  – First of all, Design & Decision Support Systems is the
    name of our Research Programme.
  – DDSS also stands for the International Research
    School, in which we collaborate with a number of
    similar groups in European universities.
  – Then, DDSS is also the name of our Master of
    Science Programme that is related to our DDSS
    Research Programme.
Design        Planning



         DDSS
Artificial Intelligence   ICT
   DDSS Research programme
• 主持人:Prof. Dr.ir. B. de Vries
• MS & PhD at the Department of
  Architecture and Building at the
  Eindhoven University of Technology
• 研究人員多達12人
• 學程包含:
   –   MSc Courses
   –   MSc Projects
   –   Graduation Projects
   –   EU and PhD projects
目前進行的計畫
             Graduation Projects
• Space utilization simulation of office buildings(空間利用模擬)
• Generative Design
• Generation of a construction planning using a 3D CAD model(3D建
  模時程規劃)
• Digitally managing the quality of (architectural and urban) designs
• Electronic Document Management in production processes
• Search systems for building product information(建築材料收尋系統)
• Digitally checking location plans
• Interactive modular house design(共同設計)
• Generating a long-term maintenance planning from product model
  data
           EU and PhD projects
• Building Management Simulation Centre
• Decision Support System for Building
  Refurbishment
• Measuring User Satisfaction through Virtual
  Environments
• Using a Virtual Environment for
  Understanding Real-World Travel Behavior
• Co-located Decision Support Space
• Simulation of Human Behavior in the Built
  Environment
• MAS for the support of Collaborative Design
      Design Systems Lab.設備
•   Desk-Cave
•   CAD software
•   VR hard/software
•   Simulation software
•   User interfaces
MAS in Collabortive Design

 Agent-mediated collaborative
 Design an building process in a
 Semantic Web context
   MAS in Collabortive Design
• 使用單一套建築輔助軟體,來協助設計師
  來滿足顧客多樣客製化需求,現已顯得捉
  襟見肘
• A system will be developed that assists
  the designer in an effortless manner to get
  information related to the current design
  task and to automatically offer solution to
  design problems.
   MAS in Collabortive Design
• The aim of this research is to analyze the
  potential of different techniques of Multi
  Agent Systems for the use in the
  domains of architectural design and the
  building process as a whole.
                             Agent                     Agent

                                     Local machine /
                     Agent   Agent      Intranet /     Agent   Agent
                                         Internet

            Human Expert                                        Human Expert
                             Agent                     Agent
     MAS in Collabortive Design
• Among the most important steps in this project are:
   – Gather information and build a knowledge base with
     minimal additional workload for the user
   – Identify problem and context based on the current actions
     of the user
   – Identify related knowledge domains and previous use
     cases, the agents representing them and the
     corresponding communication protocols including their
     ontologies
   – Gather strategies, opinions and solutions and adapt them
                       User          Suggest




                                               domain specific
                                                Add / retrive
     to the problem and hand.                                            KB




                                                  agent(s)
   – Generate suggestions and their representations and offer
                      HCI




     them in a convenient, non-distracting way
                              Listen and record
                                                              Agent society on
   – Offer approaches to user and incorporate reaction machine /
                      Generic
                                                query          local into
                                                                  Intranet /
     knowledgebase specific
                   domain
                                                                   Internet
                      application
Jakob Beetz, Bauke de Vries, Jos van Leeuwen
Design Systems group TU/Eindhoven




                         Agent-mediated collaborative
                          Design an building process in
                          a Semantic Web context
                Traditional Working Methods

• Traditional CA(A)D
  data is
  – Non-deterministic and   *


    ambiguous
                                **


  – Episodic
                                                 *


                                                 *




  – Highly dynamic                   *
                                         *




  – Does not contain
                                             *




    machine readable
    knowledge
         Central Building Information Model


• Central Building
  Information Model
  – Founded on central
    databases
  – No specification for
    interaction
  – Assumes
    completeness
Building Information Model mediated by agent technology



                                                Actor
                   Local machine / Intranet /   Agent
                           Internet




          Actor                                 Wrapper
          Agent     Agent Marketplace            Agent




                                                Resource
                                                 Agent
                                                           KB
                                                               A simple MAS scenario




                                      I would like to change the size of this room
                                            Will your HVAC unit still fit in?

                                 No                                                   Same Specs but
                                                                                     max size 2x3x4m ?

Regulations
                                                    Yes but it’s +10 dB
 Reasoner


              Sound insulation
               satisfactory?
                                                                                                          PDB



                                                                                              Yes but +10dB
Problem aspects: semantic mapping

                                         I would like to change the size of this room
                                               Will your HVAC unit still fit in?

                                    No                                                   Same Specs but
                                                                                        max size 2x3x4m ?

   Regulations
                                                       Yes but it’s +10 dB
      DB


                 Sound insulation
                  satisfactory?              Ok, we leave it unchanged
                                                                                                             PDB



                                                                                                 Yes but +10dB




                                                                                                                   Yellow pages




                                                                                                                                  A
                                                                                                                                  B
                                                                                                    UserAgent
                                                                                                    UserAgent                     C
                                                                                                                                  D


                                                                                                                                      SemWeb
                                                                                                                                      SemWeb
                 Mapping and reasoning service                                                                                        Service
                                                                                                                                      Service   PDB




                                                                                                                                      SemWeb
                                                                                                                                      SemWeb
                                                                                                                                      Service
                                                                                                                                      Service   PDB




                                                                                                                                      SemWeb
                                                                                                                                      SemWeb
                                                                                                                                      Service
                                                                                                                                      Service   PDB




                                                                                                                                      SemWeb
                                                                                                                                      SemWeb
                                                                                                                                      Service
                                                                                                                                      Service   PDB
    Problem aspects: semantic mapping


                                   SI Unit conversion Rule

   1 Energy to melt one ton of ice = 12,000 British Thermal Units per Hour (BTUH)




Cooling Unit                                    Cooling Unit
 Product X                                       Product Y



               Width      3m                                   Width      200 cm
               Height     2.5 m                                Height     422 cm
               Capacity   2 Tons                               Capacity   24.000 BTUs
               …                                               …
               ...                                             ...




                                   Mapping and reasoning service
Problem aspects: semantic mapping
                                    SI Unit conversion Rule

    1 Energy to melt one ton of ice = 12,000 British Thermal Units per Hour (BTUH)




 Cooling Unit                                    Cooling Unit
  Product X                                       Product Y



                Width      3m                                   Width      200 cm
                Height     2.5 m                                Height     422 cm
                Capacity   2 Tons                               Capacity   24.000 BTUs
                …                                               …
                ...                                             ...




                                    Mapping and reasoning service
Components of a MAS in the Semantic Web context:


                                                                                                                • Ontologies for
                                      I would like to change the size of this room
                                                                                                                  buildings, parts,
                                                                                                                  regulations…
                                            Will your HVAC unit still fit in?

                                 No                                                   Same Specs but
                                                                                     max size 2x3x4m ?

Regulations
                                                    Yes but it’s +10 dB
   DB


              Sound insulation




                                                                                                                • Mapping services
               satisfactory?              Ok, we leave it unchanged
                                                                                                          PDB



                                                                                              Yes but +10dB




                                                                                                                • Agent communication
                                                                                                                  protocols
                                                                                                                • Semantic wrappers
                                                                                                                  around Services
                                Conclusions
Conclusion:
• MAS can take care of some of tiresome
  communication overhead in distributed
  collaboration environments
• MAS in a semantic web environment can
  help to discover and process project-
  relevant information (even at design time)
• Semantic web technologies can help in a
  clean separation of Data and business
  logic
User Simulation of Space
       Utilisation
    User Simulation of Space Utilisation


• Up to now no methods for performance
  evaluation are available which involve the
  occupants of the building.
• The aim of the project is to a develop a
  method for the simulation of space
  utilisation.
  Human behaviour simulation
• Building performance analysis is a well-
  established tradition in the context of
  structural engineering and building physics.

• No model for building
  simulation involving
  the actual users.
  User Simulation of Space Utilisation

• Simulated activity
  schedule versus
  observed activity pattern.
• This project integrates
  two methods, namely
  Colored Petri Nets and
  Activity Based
  Modelling.
            System overview
Input
 The workflow of the organisation.
 The design of the building in which the
  organisation is (or will be) housed: the spatial
  conditions.
              Organisation


                               U ser
                               S imulation of
                                                Space utilisation
                               S pace
                               U tilisation


             Building design
               System overview
Output
Data about the activities of the members of the
  organisation and their location in the building space.

    From this performance indicators can be deduced, like:
   Average/maximum walking distance/time per individual.
   Number of persons per space in time.
   Evacuation time/distance.
   Usage of facilities.
   ..
                  Experiment

Using RFID to capture the real space utilisation.
Merge spaces into zones.


Compare the predicted with observed space utilisation.
Measuring Housing Preferences
     Using Virtual Reality
 and Bayesian Belief Networks
   Measuring Housing Preferences Using Virtual Reality
             and Bayesian Belief Networks
• This research aims to provide better insight in
  the housing preferences of (future) inhabitants.
  The project is guided by three research goals:
  – Develop a method (Bayesian Belief Network) to elicit
    preferences based on individually designed houses.
  – Comparison with conjoint analysis (CA) of validity and
    reliability.
  – Make a design support tool for non-designers to
    create a design.



                                       Utility Convergence
Measuring User Satisfaction
  in Virtual Environment


            Maciej A. Orzechowski
   Design System and Urban Planning Group
                   @ TU/e



        Workshop Mass Customisation
                  26.06.2003
              General Idea of
        Measuring User’s Preferences


The Virtual Environment (VE) is used to present an architectural
design to a user.


The user is asked to modify that design according to his/her
needs and desires.


Behind that visual system there is a statistical model to
estimate and predict respondent’s preferences based on applied
modifications.
               MuseV – VR System

MuseV3 – a virtual reality (VR) application with functionality
of a simple CAD system for non-designers.

Two categories of modifications:
• Structural modifications (change of layout)
• Textural modifications (change of visual impression)
           Structural Modifications

The most important from the point of view of estimation of
user’s preferences.


Change of internal and external layout



Direct impact on overall costs




Expressed in simple and direct commands:
create/resize/divide space; insert openings
            Textural Modifications

Secondary modifications (visual impact), mainly used to
check proportions, dimensions (inserting furniture) and to
decorate (applying finishes).




No influence on costs




Not included in the preference model
MuseV3 in Desktop CAVE
                  Belief Network

Searching for new, flexible method to access user’s
preferences.

Criteria:
• Interaction with the model during the time of preferences
estimation

• Possibility to find weak points (where the knowledge about
preferences is the worst)

• Improve data collection by direct feedback

• Incremental learning
             Short explanation of BN
What it is?
• Belief network (BN) also known as a Bayesian network or
probabilistic causal network
• BN captures believed relations (which may be uncertain,
stochastic, or imprecise) between a set of variables which are
relevant to some problem (e.g. coefficients and choices).

How does it work?
After the belief network is constructed, it may be applied to a
particular case. For each variable you know the value of, you
enter that value into its node as a finding (also known as
“evidence”). Then Netica does probabilistic inference to find
beliefs for all the other variables.

Incremental learning.
After the beliefs are found (post priori) MuseV updates the
network, so they become a’ priori for the next respondent.
Step 0   Step 1   Step 5   Step 15   Step 64
                     BN - Model

In our proposal the network (model) is learning  while a
user is modifying a design!


To improve the quality of collected data and the knowledge
about design attributes, the system, (based on beliefs), can
post a question to user.
4D CAD
Construction Analysis
     during the
  Design Process
   www.ddss.arch.tue.nl
     Bauke de Vries
                   4D CAD




• Linking building components with construction
  activities
• Manual task of the construction planner
• Dedicated systems: NavisWorks, 4D Suite, …
• Advantages: Simulation, Visualization
              Challenge
Automation of the planning process.

Advantages:
• Independency from the planner
• Quick first concept plan
          Implementation


                                Equipment
                                 Labour



                                  Comp.
      CAD     Constr.                        Project   Planning
CAD                               Rel. +
      model   Analysis                      Planning   Schema
                                   Dur.



              Formulas


                   Design evaluation
       Construction algoritms




Analysis by object name:
Walls are bearing floors, colums are bearing beams, etc.

Analysis by object elevation:
Object with a lower elevation is bearing an object with a higher elevation
     Construction algorithms




Analysis by directed graph:
Each object is a node in a connection graph.
      Construction algorithms

                                       G                 N
                  D
                                 F
                                                         M
                        C
                                                             L
                  B          E         J             I

              A                                          H   K

                      - Objecten zonder voorganger




Analysis by object adjacency:
Each object is a node in a topological graph
Planning comparison

                  Real
                  planning



                  Generated
                  planning
Complete process

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:0
posted:5/8/2013
language:Unknown
pages:54