Bridge Task Group Minutes
Task Group Meeting - Semmering, Austria. 2nd May 2002 [pdf download] Task Group Workshop- Toulouse, France. 3rd - 4th November 2002 [pdf download] Task Group Reference Problem Meeting - Toulouse, France. 31st March 2003 [pdf download] Task Group Reference Problem Meeting - Grenoble, France. 29th March 2004 [pdf download] Task Group Meeting - Marseilles, France. 17th September 2004 [pdf download] Task Group Meeting - Semmering, Austria. 2nd May 2002 Present: Louise Travé-Massuyès (Chair) LAAS-CNRS, Luca Console University of Torino, Marie-Odile Cordier IRISA,Philipe Dague Université Paris 13,Christophe Dousson France Télécom R&D,Carlo Alonso Gonzales Universidad de Valladolid,Michael Hofbaur Technische Universität Graz,Belarmino Pulido Universidad de Valladolid,Wolfgang Kleissl Technische Universität Graz,Mattias Krysander Linköping University Rob Milne Intelligent Applications,Mattias Nyberg Linköping University / Scania Software and Diagnostics Claudia Picardi University of Torino,Bernhard Rinner Technische Universität Graz,Iain Russell (minutes) University of Wales, Aberystwyth,Peter Struss Technische Universität München,Daniele Theseider Dupré Università del Piemonte Orientale. Apologies: George Coghill University of Wales, Aberystwyth. A. MONET - Intro MONET re-started in February 2002 under Framework Programme 5. Louise gives a quick intro to the Task Groups and lists the Objectives and also the Deliverables (listed below).   MBD1 - Special issue in IEEE SMC, session at DX / Safe Process 2003 and associated report.   MBD2 - Common Diagnostic Framework specification.   MBD3 - Report describing a collection of industrial reference problems.   MBD4 - Introductory articles and tutorials to cross-link methods and techniques to facilitate industrial   uptake of the technologies.   MBD5 - Report on evaluation of solutions to selected problems of industrial relevance. 1. MBD1 - Special issue in IEEE SMC, session at DX / Safe Process 2003 and associated report. Call for IEEE has been issued. Deadline has been extended from 1st June to mid-July. The Call will therefore be re-distributed and there will hopefully be a number of submissions, giving Bridge a good overview of the subject area as a whole. DX / Safe Process 2003. A planning meeting during DX'01 decided to co-locate DX'03 with Safe Process in Washington. Call has now gone out for Safe Process '03; DX has not yet put out its call. This co-location will be discussed in the DX'02 business meeting. 2. MBD2 - Common Diagnostic Framework specification. The deadline for completion is March 2003. It was decided that this information could be included in the first version of the Technological Roadmap (TR1), it could then be extracted to form a separate document. 3. MBD3 - Report describing a collection of industrial reference problems. The collection of Industrial reference problems is to be completed by the end of 2003. This document should possibly be completed before the Technological Roadmap but effectively the two documents will be written at the same time. 4. MBD4 - Introductory articles and tutorials to cross-link methods and techniques to facilitate industrial uptake of the technologies. This document will form a conclusion to the previous two deliverables and is due for the end of 2004 and will form an evaluation of all the three years work. 5. MBD5 - Report on evaluation of solutions to selected problems of industrial relevance. This will be an evaluation of the proposed solutions to the selected problems. 6. MBD6-7 - Technological Roadmap. A first version must be issued by May 2003. The final version for the end in December 2004. The Technological Roadmap does not have explicitly defined contents, thus these are open for discussion within prescribed bounds. Louise states that the timescale is short so we should organise the state-of-the-art section soon, this could also then be used as a separate document for industry. One part of this document could be to compare FDI with DX and the second part could be to combine or express synergies. Peter suggests that we do not want this to be solely an academic project; he makes the point that industry wants solutions not discussions. B. Bridge Membership MONET is an open Network and free to join. By the end of this session Louise would like a name added to each of the deliverables and this person to be responsible for driving the relevant deliverable, these people will be the active members. The problem is that the budget is small, only €13,200 this year. Thus it is not possible to cover the cost of everyone who wishes to attend the meetings. The goal of this meeting is to kick off the work load. In future, meetings will be based on the deliverables. Rob stated that it is not possible for MONET to cover all costs; he suggested that separate people could drive problems without the use of face-to-face meetings and thus the money could go towards a Workshop. The deliverables are of interest but some may require meetings and some may not. Louise also expressed concern over the low number of FDI people in Bridge and also the lack of industrialists. Belarmino asks if 'Control' people would qualify for Bridge, the consensus was that anyone who was solely 'control' would not qualify as a 'Diagnostic' expert and thus would not qualify for Bridge. Bridge must consider the possibility of connections with other communities - statisticians for example. Mattias N suggests more FDI people should be involved, he suggests he knows people who he could contact. The general opinion is that Marcel can help. C. Common Diagnosis Framework and Reference Problems Common Diagnostic Framework Specification could be done in a one day workshop for example. Louise wishes to discuss how we could hold a two day workshop. Louise suggests that during the proposed workshop, everyone present could introduce themselves, describe what they do and where they think Bridge should go. Peter suggests that we are a 'technology' (rather than 'application') based group and so we must form a homogeneous point of view. FDI may be more homogeneous that DX but any attempt to produce a combined methodology is doomed to failure. Thus Bridge must look at principles / aims / goals etc and then come up with high level identification of models and functions that are deliverable by our technologies (with the associated assumptions) and then add requirements and comments from industry. He also stated that this is a far off goal and may not be possible. Louise suggests that we can characterise the well recognised functions and models and when / why to use the technologies that are available. Then common grounds can be found and thus we can evaluate what different technologies are on offer. Louise is of the opinion that if all the DX participants contribute to a workshop by stating what subject areas they are looking into, then this will in itself form an illustration of the current state of the technological field. Peter suggests we could set up a format or a set of questions to find goals / inputs / gains / models etc and circulate this to Bridge members. It would also be useful to look at different technological outlooks. We must also be prepared to "sell" these goals and this requires openness in Bridge. How are we going to collect Industrial Reference problems? Rob states that (with respect to Gas Turbines) the US Government / Navy / Industry is very pro MBD, however he states that to them it meant 'numerical' modelling. Rob states that the major challenge is to convince these groups otherwise. Louise noticed that in Europe the situation may be different as DX MBB technologies have been supported through several european projects. Rob also volunteered to provide data on this area. Peter states that we have to be aware of the 'application' of technologies not the 'areas', i.e. we must look at technology not specific problems. Therefore we must identify the characterisation of the tasks / observables etc. First step is therefore to produce a Table for dimensions of these technologies. Louise has one available that can be altered and extended for the purposes of Bridge. ACTION: This Table will be circulated (in a paper) for people to add dimensions to, they must add in how they feel about the subject areas and why. Another problem is that Industrial Relevance requires problems of "How" is something solved, i.e. "this is the problem, what do we do now?" First MONET must get industrial people interested. Louise knows of bodies with actions going forward and they could be approached by Bridge. In this way we could make a list of other European actions, CHEM is one example. One problem will be that industrialists can often not give up the required research time and therefore we must begin by inviting them to provide feedback instead of requiring an active part. After reference problems are collected these could be given to industry, for comment. Rob suggests that we should include people like Peter and himself, i.e. suppliers to industry rather than the end users themselves. Therefore we can look at 'industry' as a two layer target. Peter could also assist with industrial contacts and reconnect the MONET1 contacts in Rotterdam (Unilever and others) who meet to discuss these issues. ACTION: Louise will contact 'Damadics' Network as they are working on similar areas. Peter to attempt to re-contact Rotterdam institutions. With regard to the Workshop, it needs to be after the 'Table' has been circulated and the first selection of reference problems has been collected. Thus a date of September of October is suggested. Also descriptions of technologies and reference problems need to be collected. An attempt was made at DX'97 to produce some industrial reference problems and this failed, it is suggested that this was due to the efforts being closer to a 'Bench Marking' exercise. So to avoid similar problems we need to focus on areas that can be achieved and also on problems which industry can work to a solution as well. Basically we must define what we are after, first we could look at FDI domain and show how DX could do it better or go beyond what FDI is capable. However we must ensure that our problems are practical and we must have solutions which are applicable. Deadlines Workshop - end of October / November 2002 Table finalised by end June, to be circulated immediately. Format for description of reference problems for end of June, to be circulated immediately. Information is to be collected by the end of September. The question was raised as to whether we are after reference problems or descriptions and standards as well, it was agreed that maybe this should wait until after the workshop. However if we specify the reference problems this could do both. Thus we should look at reference problems and underlying assumptions, etc. Thus any submission to this work should include a breakdown. This has been done by Louise, Marie-Odile, Philippe and colleagues within the ALARM French Group, they have written a paper that includes a Table that Bridge could use. Action: This paper is to be circulated (which includes both a table for dimensions of the MBD technologies and a format for the description of reference problems, from which we can start). D. Dissemination of Materials and Tutorials The bridge Workshop at DX'01 had papers on this topic and these are still accessible online. We can make a list of recent relevant papers (submissions to the IEEE SMC Bridge special for example). This list should then be structured for target audiences, both academic and industrial and be themed, i.e.: theory, applications, etc. The question was raised as to whether Bridge wish to participate in the MONET Summer School? What contributions does Bridge want to make? Should there be a presentation on a BRIDGE topic? Uniform positive response was received it was generally felt that this could be a presentation of the Common Diagnosis Framework. Action: Louise to initiate the list of reference papers. E. Preparation of Technical Roadmap This will link in with the FP6 preparation - dedicate 3 hours of the workshop for a Technical Roadmap brainstorm. F. Preparation of FP6 Expression of Interest This must be formed correctly. The idea is to show interest and attempt to get Bridge included in FP6 directions. A questionnaire asking for views of technologies has been sent out and answers should be returned by 12th May. Action: send Questionnaire to Mattias Nyberg - (done 8/5/02). Peter states that we should look at exactly what the Commission expects. Automotive Task Group will look more to an integrated Project Approach, whereas Bridge is probably more Network based. Action: speak to Knud (our Commission Project Officer) re: the Expression of Interest, as Luca suggests this may be a way into Europe. Then possibly meet Knud in Brussels at the end of May. Also Marcel Staroswiecki is preparing a proposal for a Network of Excellence called "Dedals" (DEsign of DependAble and reLiable Systemsems). He aims at putting together Control / AI / Signal communities. The Bridge topics are hence included in those of Dedals. Task Group Workshop - Toulouse, France. 3rd - 4th November 2002 Details of the workshop are on a separate page. The minutes of the meeting are available here in the usual format. Bridge CDF and Reference Problems Workshop minutes 25 – 26/11/02 at LAAS-CNRS, Toulouse, France Present: Louise Travé - Massuyès LAAS-CNRS, Joaquim Armengol Llobet University of Girona, Teresa Escobet Canal University of Catalunya, Janet Thomas MONET, Rob Milne Sermatech Intelligent Applications Carlos Alonso University of Valladolid, Belarmino Pulido University of Valladolid, Anne-Lise Gehin LAIL, Lille, Tatiana Kempowsky LAAS-CNRS, Toulouse, Gilles Dumont D’ayot LAAS-CNRS, Toulouse, Mattias Nyberg University of Linköping and Scania, Olivier De Mouzon LAAS-CNRS, Toulouse, Philippe Dague University Paris 13, Daniele Theseider Dupre University of Piemonte Orientale, Benny Raphael EPFL, Xavier Olive ACTIA/LAAS-CNRS, Toulouse, Hervé Poulard ACTIA, Toulouse, Stéphane Ploix Laboratoire d’Automatique de Grenoble (LAG), Samir Touaf Laboratoire d’Automatique de Grenoble (LAG) Peter Struss OCC’M and Technische University of Munich. Monday 25/11/02 Louise Travé Massuyès welcomed everyone to the Workshop. She stated that it would focus on the Common Diagnosis Framework(CDF) and on Industrial Reference Problems. One of the tasks of the Bridge Task Group is to deliver a document proposing a CDF and another presenting a selection of Reference Problems. It was decided to get a good illustration of the problems by starting with the views of the industrialists and technologists to obtain their needs and requirements for integrating the technologies on both sides or using the separate domains. Louise then gave a brief overview of the MONET Project and explained the Task Group’s deliverables. Industrial view and CFD sessions There followed a series of presentations from some of the industrialists present: Rob Milne, Hervé Poulard and Peter Struss. This was followed by a discussion session, reported below. After the industrialist’s view the research view was discussed. In the research world, one is looking to create revolutions! This was a contradiction of one of the industrialist’s claims during the presentation. There are two differing viewpoints. Therefore it was decided to try to define a method of working. First to define a starting point, then decide where the discussion was aiming. From an FDI (Fault Detection and Isolation) perspective there are many different answers even to the question of what FDI1 is! So there is a need to determine which FDI task(s) we are looking at.It was agreed that we needed to make our assumptions explicit and definite. Indeed, there are many implicit assumptions in FDI. It is hence difficult to compare the differences. The FDI community does not want to prove the same things as the AI community. It was agreed that AI and FDI have different goals which explain the different approaches. It was then suggested that we look at some slides to do some comparison and the discussion followed. This discussion will provide several elements for the CDF and will be reported in the corresponding report deliverable. Tuesday 26/11/02 CDF, taxonomy session The morning started with the presentation some slides by Stéphane Ploix, presenting the application of a typical FDI approach and of the DX approach to an academic example. The discussion about the CDF continued. There were then two presentation about taxonomy (The ALARM Group taxonomy by Louise Travé-Massuyès (see Tables of Dimension) and the University of Valladolid supervision taxonomy by C. Alonso which gave rise to more discussion about the CDF. The relevant elements will be put together in the corresponding CDF report deliverable. Reference Problems Session There has to be an agreed format to the problems. It was decided that the reference problems would be restricted to the strict FDI tasks. However, relevancy of the FDI outputs to input other supervision tasks (i.e. fault accomodation, reconfiguration, etc.) would be mentioned. 1 FDI may refer to the tasks “Fault Detection and Isolation” or to the FDI community. The meaning is generally clear from the sentence context. A format was agreed for the reference problems as follows. A. Problem Description 1. Monitored Physical System • Specifications / Requirements • Kind and level of complexity 2. Monitoring aims / Supervision Problem 3. Desired outputs from the FDI System 4. Available Knowledge 5. Available Data B. Existing Solutions 1. Inputs 2. Outputs 3. Models 4. Relevancy of the FDI outputs to input other supervision tasks (i.e. fault accomodation, reconfiguration, etc.) 5. Implementation 6. Validation, Maintenance, Evolution Some Reference problems were then presented to the workshop. Name Problem Type of Problem Peter Struss Large Matrix switches Pure isolation problem Anne-Lise Gehin Steam Generator Pilot Process Process Control problem Teresa Escobet Damadics Process Control faults, sensor faults Carlos Alonso Sugar Beet Process Process Control faults Olivier De Mouzon SCAO Hybrid detection, isolation Benny Raphael Modelling A Bridge Distributed parameter models Rob Milne Gas Turbines Condition monitoring, alarm filtering FP6 Issue Louise reported about the Bridge EoI and the feedback received from the MONET officer. It appears quite clearly that the Bridge topic is too limited for a NoE and that its vertical nature is not appropriate for an IP. Louise suggested that Bridge should consider becoming a part of the DEDALS Network, whose EoI was proposed by Marcel Staroswiecki (LAIL, France) for FP6, working under the MONET umbrella. Marcel can provide more information about the current status of the project. Bridge can also be relevant in several more horizontal IPs. Bridge members are encouraged to participate in such proposals, and to consider the other Bridge Task Group members as possible contributors if needed.   Task Group Reference Problem Meeting - Toulouse, France. 31st March 2003 Present: Louise Trave-Massuyes, Belarmino Pulido, Carlos Alonso, Jose Aguilera, Xavier Olive, Rob Milne, Teresa Escobet, Anne-Lise Gehin, Benny Raphael, Miguel Gonzales, Janet Thomas. Agenda Reference Problems Reference Problems Louise has put together the reference problems received so far. They are not all available yet. From the last task group meeting there were 7 problems. There is documentation on 4 of them and paperwork to follow for the others. The plan for the afternoon is to review the reference problems. 15 minutes for everyone to read the documentation and then a discussion and analysis of them. The idea is to analyse which types of problems are most relevant and why. We need to show why Bridge is either relevant or not to the different application domains. For example, Peter Struss' example is relevant to DX and not FDI. The problems types were then summarised and the existing solution techniques were explained, including why they were chosen. Reference Problem 1 - Anne-Lise Gehin. This problem calls for different types of model - all complementary. It is now approached using bond graphs, which are usually used to derive the equations, which are highly non linear in the present case. A constraint based technology could also have good potential. What type of problem does this represent : the difficulty is in the non linear aspects, but since it is a pilot plant, there are no noise, neither disturbance problems. The pilot is equiped with redundant components so as to allow for reconfiguration. Few real systems offer the opportunity to reconfigure and this is a real advantage. This is a lab situation, so there is the possibility to know what faults can be introduced to system! Therefore one can focus on the known faults, which is never the case in real world processes. Not usually so clean an environment. Several approaches will be tested - unsure yet which ones - within the framework of the CHEM european project. Patrick Talibert will possibly use a compiled approach using the expression of residuals. It may be interesting to see a logic engine based approach. Bridge can help on the falt isolation side Reference Problem 2 - Rob Milne Gas turbines are noisy system with bad sensors. There is lots of data, so it is often difficult to isolate the exact problem. In this type of application several types of approaches are required - rule based, FDI and neural networks. FDI uses dynamic sequence events. Often these are too high or low and then a rule based application is needed to diagnose the sequence of events. TIGER is a hierarchical system, which has 600 inputs per second. It therefore does continuous fault detection and diagnosis. Model Based techniques are used on the fuel valves, which are noisy, so lack some sensor readings. Neural networks worked very well where the input conditions are stable. They were applied on on sub-systems successfully but many functions are not stable enough and have many variables. The technology works very well, but inputs are not stable enough. Rule based methods are therefore better for broad ranges, while MBS is better for smaller, more specific systems. Without rules, it couldn't do top level decisions with the variety of inputs and faults. FDI could not cope with the complexity. Models would be too complex to represent the whole system. Some statistical techniques have been tried as well. Fault isolation could work, but the scale is very large with 35 different turbines running, of six major types. The main problem is with signal analysis, because signals are noisy and there are many unexpected disturbances. Complex and many faults occurring at any one time. The set of available sensors could be improved to get better redundancy. TIGER mainly uses raw data, but event detectors do some filtering. The aim is to give the owner information on what the main problems are on the turbine, so that when the turbine is shut down, essential maintenance can be carried out based on the information supplied. This shows up the symptoms, which have to be identified by the operator. The system doesn't always do diagnosis, just detection. In engineering, we have usually been working with end users, not designers. Reference problem 3 - Carlos Alonso & Belarmino Pulido The system is so complex that it needs to be decomposed and therefore a Model Based approach is appropriate on the sub-systems and FDI approach can be applied with expert knowledge.W e also want to include machine learning for pattern recognition. There are similar problems to those encountered by the gas turbines of TIGER (they are all connected industrial steam processes!). There are possible structural faults - may be output erroneously corrected. Would qualitative models work? Only a finite number of models? If only one type of model to be used, it would have to be numerically accurate. But a combination of quantitative and qualitative seems the most appropriate The real problem is that exact modelling is impossible. Reference Problem 4 - Teresa Escobet This problem is classified as identifying the different faults. Several methods are being applied in the framework of the DAMADICS european network: Bond Graphs, parity equations, … and quantised systems and non-linear models. Approaches arre mainly FDI techniques, not DX. As a matter of fact, there may be more problems for FDI and not DX. The conclusions of the DAMADICS project are not yet finalised. Fault models were built for some problems but some physical phenomena, like friction, cannot be estimated easily. In the last two years there has been a different focus - isolation and identification. FDI will just indicate structural impact on normal behaviour model. DX really needs fault predictive models. END OF FIRST DAY Reference problem 5 - Benny Raphael This reference problem concerns structural diagnosis of bridges. It is approached by static fragment modelling. Hypothesized fragments are assembled and parameter estimation techniques are used to perform model updating. The are people working on vibration information, but in the present case, it is difficult to perform vibration tests. The model is highly non-linear. Liliana Ironi is working on similar model identification problems - how to use qualitative simulation to fit the structure of the model. Ben Kuipers has also recently published similar work (last month's AI Magazine ?). There are too many parameters - often different values can give identical results, so more information is required to determine faults or if the correct model has been used. there are also have noise disturbances - these are mainly filtered out manually. It is not easy to establish a baseline of readings and original inout variables. This problem is as interesting as atypical. It seems that no other technique than model updating could be used, given the constraints. Reference Problem 6 - Olivier de Mouzon & Louise Travé-Massuyès This reference problems is concerned with the attitude and orbit control system of a satellite. This is a closed loop system with few measurements. Another difficulty is that we have to cope with different types of components, continuous as well as discret event. The satellite sensors are subjected to noise and the satellite is subjected to unknwn disturbance. The ultimate goal is to perform fault-tolerant control in the sense that state identification (diagnosis) is used to trigger the reconfiguration of the system, so that the satellite recovers a normal state or operates under degradated but safe conditions. The existing solution uses hybrid systems, composed of probabilised automata and Ca-En type models for the continuous aspects. Reference Problem 7 - Peter Struss Ask Peter to provide input.   Bridge Task Group Meeting - Grenoble, France. 29th March 2004 Present: Louise Travé-Massuyès LAAS CNRF Robert Milne IA Teresa Escobet UPC Peter Struss OCC'M and TUM   Belarmino Pulido University of Valladolid Carlos Alonso University of Valladolid Marcel Staroswiecki Lille Marie-Odile Cordier IRISA Joaquim Armengol University of Girona Cecilia Zanni LSIS Fabrice Auzanneau Commissariat a l'Energie Atomique Stéphane Ploix INPG Samir Touaf INPG Iain Russell MONET Janet Thomas MONET Apologies: Philippe Dague, LIPN, Paris 13 Benny Raphael, EPFL, Switzerland Agenda   Reference Problems   Roadmap   FP6 and EU Issues   AOB. The main purpose of this meeting was to discuss the reference problems and their solutions. The proposers of the reference problems gave a short summary and a ten to fifteen minute presentation on the solutions and the group then discussed synergies of interest but as yet undeveloped. Reference problems   CHEM - Marcel Staroswiecki   Sugar Beet System - Carlos Alonso   Gas Turbine - Rob Milne   DAMADICS - Teresa Escobet   AOCS: Attitude and Orbit Control System - Louise Travé- Massuyès   Bridge Information - Benny Raphael (Could not be present but sent an Article and a detailed account   which was read by Louise Travé- Massuyès and discussed and handouts issued for later feedback.) Discussion MBD5 - a Report on Evaluation of Solutions is due August 2004. Marcel will make CHEM Members only area of the website available to MONET Members. Sylvie Cauvin will talk at DX'04. The steam generator is available (and other models) on the web site. The idea of writing a collective paper with the information collected for MBD3 and MBD5 was mentioned but it was agreed that many papers had already been published about the considered reference problems. Many of the reference problems are in the same class. We should look at including problems that contain discrete event approaches. Are there any difficulties if the paper is for the DX community, or for the wider AI community? We need to decide what the major message is and who the target audience is. We should identify areas which need to be investigated further and which gaps to fill. The real objective is to get industry to understand where our techniques apply. We need to look at complementary, as we are mapping different approaches. What we can achieve at this stage is to look at what approaches have been used etc. The situation has been static for a while, so a clear view is needed. There are different approaches as the difficulties are at different levels. It may be premature to consider a paper at this stage. Comparison of detection methods is required. Stéphane will provide some references to Louise. We should also consider another work from the 1998 DX by Mike Chantler and investigate whether we have a better understanding now than we did then. We have a better view on how Control and AI methods bridge, but not many other advances. This does not help in classifying the other methods. Conclusion of reference problem section We need the presenters to provide the details of their reference problems for the deliverable. This should take the form of an extended reference problem description (based on that produced for MBD3 Existing Solution Section) showing what has been presented today and summarising the conclusions. Louise will write an introduction to the document - a summary of the current state-of-the-art/technology. This will be submitted to the Task Group for feedback and then we shall collaborate on the rest of the document.   Teresa will submit information on DAMADICS by the end April,   Benny will provide an editorial on his bridge problem,   Marcel will provide information by the end of May from CHEM,   Louise will provide introduction for feedback by the end of May There are characteristics that are missing. We would need a pure discrete event example to complete the list. System plan Characterising the task - detection/not detection… Subtests The view here is to start with the reference problems, and then consider the task is required for every reference problem, e.g.: steam generator only dealing with internal fault diagnostics, but could have easily looked at supervision as in the sugar plant example. The DAMADICS data and actuator model is available, but experiments may no longer be possible. We need to look at what is available - Model, simulator and historical data will be available. There are also links to the DAMADICS benchmarks information which could be very useful. We should publicise that these are available and that people are welcome to use them via our website. AP1: Janet to check what will be available from DAMADICS. AP2: Janet to look into placing these links on the bridge pages. Road maps Iain briefly outlined the existing Roadmap and the process required to update it. Some updates and amendments to the current road map were discussed and these will be incorporated into the next version of the Roadmap. A possible date for a road mapping meeting to finalise them and add the rest of these amendments on such things as the environment and other issues could possibly the 5th - 6th July potentially in Marseille or Gerona or Paris or Barcelona. AP3: Iain to provide Road Map information as above for inclusion (or prepare new document?) FP6 and EU Issues Due to time constraints of train and flight times, these issues were agreed to be sent out for people to read on their own time. AP4: Iain to e-mail FP6 and the next funding information to the Task Group. AOB The next Task Group Meeting will be combined with the Road Mapping Meeting, to be arranged for early July. Meeting closed 4pm. Action Points AP1 Janet to check what will be available from DAMADICS. AP2 Janet to look into placing these links on the bridge pages. AP3 Iain to provide Road Map information as above for inclusion (or prepare new document?) AP4 Iain to e-mail FP6 and the next funding information to the Task Group. Bridge Task Group Meeting,Marseilles, France. 17th September 2004 Agenda 1. Roadmap 1.1 Presentation of Roadmap 1.2 Possible Improvements   Discussion of Drivers   Drivers   Technologies   Products 1.3 Specific topics   Areas not covered by Roadmap 2. Taking the community forward - needs and directions 2.1 Motivation of Researchers 2.2 Wish-list for Diagnosis 2.3 DX/FDI Convergence Attendees Joaquim Armengol Marie-Odile Cordier Teresa Escobet Marc le Goc Micheal Hofbaur Rob Milne Belarmnino Pulido Benny Raphael Iain Russell Sandro Saita Louise Travé-Massuyès Marina Zanella Cecillia Zanni Minutes Introduction by Louise Presentation of RM Discussion of Modelling Issues  (Action) MH to send info on MOBIS Project Discussion of Temporal issues Autonomy vs Human Interface in Applications Graphical RM - Technologies Taking the Community forward / direction and motivation DX stream at SafeProcess was a good idea, not necessarily having the co-location. Two communities have been evolving in the awareness of each other, much of this was due to the efforts of MONET. Now the Bridge is established it is through the Science that the development should be pushed, we have possibly got as far as basic awareness has been done. It takes a lot of effort for a researcher to absorb another approach, but we should look into focusing PhD students on this area (hybrid diagnosis for example) and will therefore grow up in both technologies. If we keep MONET electronic newsletter going we should make an effort to involve SafeProcess People. Is it useful to have a workshop that is on all the approaches to (automated) Diagnosis (dx, fdi, neural networks, expert systems)? This could encourage a wider basis for DX Workshops. Could this be a special Issue of a Journal? Bridge Special (IEEESMC) Issue is to be published in October. Also we could try to have a model-based / Diagnostic element to ETAI, the papers are put up on the web and reviewed by anyone, but certain people have review up on web. If it is accepted it publication date is the original date it went on the web. Benny reviews for Advanced Engineering Informatics, and if anyone has an engineering paper they can send it to him. Also we publish papers in a variety of journals, what we could do is to locate the papers and point to them on web. (Action) Iain - email out for this. Products Drivers
[Page top] Any problems with this page please contact: Web maintainer Page last updated: January 4th, 2005