KDI: Networked Engineering A Joint Research Initiative of CMU-Drexel-USC William C. Regli Assistant Professor and Director Geometric and Intelligent Computing Laboratory Department of Mathematics and Computer Science Drexel University http://gicl.mcs.drexel.edu KDI:NE Project Goals To develop, integrate and evaluate information systems for distributed and collaborative design and manufacturing. Project Overview • NSF Grant: • Program: • Amount: • Duration: CISE/IIS-9873005 Drexel University NSF Knowledge and Distributed Intelligence (KDI) Initiative $1.2M October 1998 -- October 2001 One of 40 projects selected in 1998 (of 697 proposed) Principal Investigators • Carnegie Mellon University – Pradeep Khosla – Ramayya Krishnan • University of Southern California – Berok Khoshnevis – Stephen Lu • Drexel University – Thomas Hewett – William Regli Industry and Gov’t Partners • AT&T Labs, Internet Platforms • National Institute of Standards and Technology • Structural Dynamics Research Corporation (SDRC) • Delaware Valley Industrial Resource Center (DVIRC) • Bridgeport Machine Tools • Ford Motor Company KDI: Networked Engineering • Communication of Information – within an office – across virtual organizations – to suppliers and customers Information • Access to Services – human expertise – software agents Services Collaboration • Collaboration & Negotiation – among different disciplines and departments KDI:NE Research Objectives • Design Repositories – Engineering knowledge-bases to leverage legacy knowledge • Composable Simulation – CAD enhanced with engineering analysis and behavior models • Collaborative Negotiation – Conflict management strategies for design • Usability Evaluation – Assess computational support for collaborative design • Networked Engineering Studios – Integrate Internet, collaboration, and CAD tools Design Repositories Goals: Record, archive and manage design information as it is created during distributed design activities. Approach: – Message model for distributed design – Process archival methodologies to populate design knowledge-bases – Retrieval strategies for design knowledge-bases Impact: – Enables variational design – Access and reuse of legacy data and information – Platform for networked collaboration on and knowledge sharing about design problems Related Project: NSF CAREER Award CISE/IIS-9733545 Repository Scenario Engineering Digital Libraries s M(θ) θ V(θ, θ ) G(θ) F(θ, θ ) … contain CAD models, assemblies, plans, revisions, S-B-F models, project information and workflows, design rationale... Research Objectives • Integrated engineering knowledge-bases and engineering digital libraries • Intelligent decision support tools for design • Techniques to leverage legacy knowledge Current Results and Accomplishments: • Graph-based structures for knowledge modeling • Geometric search algorithms • Collaborative/Conceptual interfaces • Internet-Based Design Repository National Design Repository http://www.parts.nist.gov http://repos.mcs.drexel.edu • Enables national and international participation • Links in related resources • To be coupled with intelligent search and retrieve tools Involved Drexel/GICL Research Personnel • • • • • • Dr. William C. Regli (MCS) • NSF REUs – Lisa Anthony Dr. Thomas Hewett (PSA) – Dmitry Genzel Dr. Wei Sun (Mech. Eng.) – J. Elvis John Dr. Jon Sevy (GICL Asoc. Dir.) – David McWherter Mr. Gaylord Holder – Yuriy Shapirshteyn GRAs – Joshua Wharton – Vincent Cicirello (MS, 1999) • Part-Time & Workstudy – – – – Xiaochun Hu (PhD) Max Peysakhov (MS, 2000) Xiaoli Qin (MS, 1999) Vera Zaychik (MS, 2000) – Binh Le – Victoria Charles Current Status • Deploying Repository site • Initial implementation of – Conceptual query interface – Structure matcher • Data acquisition – CAD data (w/ SDRC, PTC, NIST) – Process/Assembly Plans (w/ Bridgeport and CMU) • Integrated with fabrication services – GICL’s Bridgeport 4-axis machining center Future Work • Integrate Repository with K-base system • Approximation algorithms for – structure matching – distance measurement • Enhanced design graph representation • Experimentation and testing of conceptual design/query interface • Adaptable query interface for Internet agents • Integrated cost estimation, planning, and manufacturing network services Composable Simulation Goals: Create simulations of mechatronic systems by composing mechanical CAD models, electrical models and information technology. Approach: – – – – Automatically create product-level simulations CAD enhanced with analysis and behavior models Hierarchical distributed simulation architecture CORBA-based implementation Impact: – Allow reuse of simulation models – Significantly reduce the time to build simulators – Increase fidelity of simulations Scenario: Conceptual Design... p p Pitch s m Control Signal Reference PID s K s ( m p ) m J m bm s Coupling Pitch Motor s M(θ) θ V(θ, θ ) G(θ) F(θ, θ ) PID y s m Control Signal Reference Mechanical System s K s ( m y ) m J m bm s Yaw Motor Coupling Yaw y … to Model Simulation Design concept CAD and Virtual prototyping Ref. Control Yaw motor Dynamics Control Pitch motor Automatically generate dynamic model and simulation software. Ref. System Overview Conceptual Design Component Models Simulation software architecture Linpack Simulation processes Odepack Matlab Information Agents Dymola ACIS Novel Features • Creation of simulation software by combining individual simulation processes • Inclusion of information agents in simulation process • Provision of distributed environment • Automatic model refinement Component Models • Object-Oriented Modeling Paradigm – reuse of models • Design Repository used to select CAD components – incorporates ADAMS or DADS • Information Agents – control system algorithms – environment definition Software Architecture • Analyze conceptual graph to create simulation processes – distributed objects – retrieve CAD information via ACIS • Build simulator architecture – synchronization mechanisms – communication protocols • Execute simulation Simulation Output Collaborative Negotiation Goal: Develop systematic methods to establish optimal strategies to guide design team interactions and to manage design conflicts raised from these interactions. • Approach: – Game-theoretic modeling of collaborative design activities – Establishment of conflict management strategies for mechatronic design problems • Impact: – Theoretical foundations for new software tools to support collaboration and negotiation activities – Techniques for trade-off analysis in mechatronic systems design • Related Projects: – DARPA/CMU CODES – CMU DecisionNet Outline • The problem context – enterprise-wide decision support for military logistics planning • The approach – use an e-commerce metaphor to create a virtual repository of decision support resources • The research challenges – metadata (what kinds, representation..) – the search and discovery problem Overview of DecisionNet • Architecture – Providers/developers of decision support objects • register with broker • provide metadata – Broker(s) • compiles metadata into a catalog • supports search to respond to consumer queries with varying degrees sophistication • returns executable plans (an ordered collection of services) – Users • use broker to search and retrieve resources/computational plans • use broker to execute resources to solve problem • computational objects in the repository Networked Engineering Studios Goal: Deploy testbed of Design Studios that integrate Internet technologies and collaboration/multimedia tools with proactive CAD systems and inter-networked engineering and fabrication services. • Approach: – Leverage vBNS, COTS software and strong industry collaboration – Merge collaborative work tools with tools for design, manufacturing, negotiation and Product Data Management (PDM) • Impact: – Interdisciplinary learning/work environment: CS/EE/ME/Psyc/CE/IE/HCI – New classroom for industry – Platform for evaluation Initial Sites vBNS Logical Network Map 75 Operational Connections 19 Planned Connections Last Updated 2/1/99 Wisconsin @ Madison Wisconsin @ Milwaukee FNAL UIC ANL Chicago Northwestern Dartmouth Notre Dame Minnesota Iowa State UNH Brown Boston U Purdue Washingto n MIT Harvard Indiana Iowa MREN/ STARTAP NREN PNW DREN ESnet Rensselaer Rochester CMU Merit Penn State Michigan UC Boulder PSC Michigan State NREN Boston Syracuse NYSERNET Cornell Ohio State NASA AMES NCAR UNM SUNY Buffalo Wayne State DREN Seattle Yale SREN NI Oregon State UMass CA*Net II APAN ESnet NYU Rutgers Utah Chicago Columbia Cleveland Princeton New York City UC Davis DREN UC Berkeley Denver UCSF NCSA Missouri UIUC Stanford CalREN-2 North UCSC iDREN Sprint NY NAP Washington @ St. Louis Drexel UPenn Perryman, MD Johns Hopkins San Francisco UMBC UMD UCLA CalTech UCSB USC Washington DC Cal Poly Pomona SDSC UC Irvine UC Riverside Cal State San Bernardino Kentucky vBNS Approved Institution Planned vBNS Approved Institution vBNS Partner Institution Network of vBNS Partner Institutions 1998 Planned Network of vBNS Partner Institutions Texas Aggregation Point Planned Aggregation Point DS3 OC3 OC12 DREN NIH NCSC Vanderbilt SDSU ESnet iDREN MCI - vBNS POP Wake Forest UCSD NI MCI Reston Houston Arizona MAX UNC UT Austin Highway 1 Alabama @ Birmingham Duke SoX VA Tech ODU Baylor C. of Medicine IB&T @ Houston Houston Rice TAMU MFS DC NAP NREN Atlanta CalREN-2 South USC ISI ESnet Los Angeles GA State NC State GA Tech UVA George Washington Florida FSU Miami Georgetown Central Florida NOTE: Lines between institutions and aggregation points or NAPs represent the configured bandwidth of their connection to the vBNS. The bandwidth of the actual circuits may be greater than shown. Evaluation & Human Factors Goal: Work with practicing engineers and engineering educators to improve support for design and to understand performance of designers in distributed and collaborative design environments. Approach: • Assess computing system support for design and collaborative design through empirical examination of interaction effects among: – Hardware and Software characteristics – Identifiable sets of users • Evaluation as part of computing system design process: – – – – Provides feedback to designers Enables users to contribute tool and system design ideas Forces ongoing concern with goals and the criteria for meeting them Evaluation will happen Impact • Improved tools for engineering design and collaboration • Tools that designers will want to use • Assessment techniques Collaborators: – – – – – – DVIRC Drexel ME SDRC Bridgeport Ford NIST Assembly Structures CUP: Conceptual Understanding and Prototyping • Functional requirements “Enable back-of-the-envelope sketching” – capture basic 3D assembly structure – part relationships – function characteristics – behavior characteristics • Collaborative environment • Internet-Centric