STARS students, STARS coaches, professor Bernd, professor Scherlis, Jane Siegel, Dick Martin, Tom Hawley
Clients:
Chuck Dipman from Raytheon, Los Angeles
Chip Morris from ManTech, West Virginia
Bob Wonders from MTS Technologies, Johnstown, PA
Allen Ellmaker from Wavecom.
Time:
9:30-10:20a.m.
Questions and Answers
System Design and System Decomposition
Speaker: Dan Heller
At the end of the presentation
Jane:Why not consider pen-based input for maintenance control? Dan:The current system design does not rule out open-based input as a possiblity. In order to use the utensil for a pen-based system, the inspector or repair person needs one or both hands free. Therefore, I am not sure if pen-based system would add more flexibility to the maintenance operations. Bernd:Yes, as Dan has remarked, we can use pen-based system for inspection tasks. We are not ruling out pen-based system as an option. However, we are not focusing on it at the moment.
Bernd:why are there no hardware requirements for inspection? Shouldn't wearable computers be included here? Li:The wearable computer is not specific only to inspection. Since it is not owned by inspection, we did not put it there. Bernd:Why isn't ACS on there? Li:We have heard that STARS would no longer use ACS
Bernd:When you say things are going to be efficient, what do you mean exactly by that? Li:It would mean run-time efficiency Bernd:So are we going to talk about the Modeling Algorithm? Paul Shao (from modeling):Yes, I will address that in my Data Management presentation later on this morning Dick:Just for your information, we have the Lucent Radio LAN cards. We do have some IBM laptops that perhaps can be utilized to test the system.We can probably configure those labtops so that we are testing on a squeezed-down model of it with the exact memory requirement as the system that the clients are going to be using. Bernd:What are we going to be using for serial connection? Li:We are deciding on using USB ports Bernd:Is this decision based on a thorough review of all the choices? What is the reason that you choose USB? Li:This choice is arbitrary. It is decided upon by the Augmented Reality subsystem. Li:We would also like to find out more about the GPS equipment from the client.
(No clear answer about GPS from the clients)
Dick:I just want to mention that you have to be careful of serial I/O. This is because there are some issues with configurations with serial connections that things might not always work.
Bernd:Is it possible to have a conflict with annotation of IETMs? Dick:Putting annotations on IETM documents should be more of a serial process. There shouldn't be any conflict with regard to annotations. Li:I am sure someone from Authoring would be able to address this question more appropriately. Jon (from Authoring):Yes, it is indeed possible for annotation conflicts once the IETM document is out on the field because everyone can edit it at this point.
Dick:Is there a concurrency problem for maintenance operation? I don't think so... Jane:No, the maintenance operation would be a serial process.It is possible to have more than one person working on an aircraft. But each of them would have a separate work order to work with. Bernd:In order words, the relationship between an inspector and a work order is one to one. Jane:Well, is there a mechanism to prevent a work order being accessed by more than one person? Li:Yes, each person who is working on a work order would try to obtain a lock on it. Once someone has the lock, no one can access that work order until that person releases it.
Dick: how much processing do the model need? What about disk space and speed? I mentioned that because we have to be realistic about this since our last try at this choke the old processor and it just didn't work. Paul: we have a very efficient algorithm for this purpose. Bernd: so you just download things as you need them Paul: yes, we only fetch things as they are needed. Dick: you need to look at the speed of the radio LAN to see if it can handle what you are describing. Paul: well, this is more like we plug in and get it. (not clear)
Dick: You say that you are using Lotus Notes as your Database. But I think you should consider other databases as well. Paul: Yes, but it wouldn't matter which database we choose because we are treating it as a black box and all the subsystems would be interfacing with it through the API that is provided by workflow.
Bernd: You said the IETM size is 80 mega bytes? I really don't think that is the right size we are talking about here because it is not efficient to be downloading 80 mega bytes of IETM on each increment. Jon (from Authoring): well, the size of IETM sections that the system is going to be downloading would actually be much smaller. We call this smaller entity nodes. Bernd: we want to make sure we can do incremental downloads efficiently. Jon: Yes, we are striving toward that goal.
Bernd: can you give us some numbers for the size of the models we are storing in the PEDD. Paul: Currently the model of the plane takes about 3 mega bytes. The size of the model is really depended on how deep we go into the hierarchy for the model of the plane that we are storing. If we do it up to 3 levels down it will take about 20 mega bytes. Dick: For the speech recognizer, there are a few issues that we need to address here. The first is we might want to put some constraint on the number of vocabulary it recognize since it will just be a big hog for the system if we don't think about this realistically. Second, we might want to think about the coordination issues between the user and the speech system. Some of these things include the ability for the user to go back to the previous instruction or page or maybe ask the system to repeat the same instruction again. Jason (from repair): Yes, actually we have been thinking about that. The way we are going to address the coordinate issues is to have some keywords that the system will recognize to perform specific tasks. Examples of these would be for the user to say "next step" when asking for the next instruction or say "repeat" for repeating the same instruction again.
Dick: we should have the maintenance control going to fill out the work order. Steve: ok Dick: We have to have some scenario to demonstrate how work order works Steve: ok
Bernd: We should not name it workflow database. But rather it should be named as STARS database or some other appropriate name, because the database does not belong to workflow only. Steve: ok
Dick: Are you guys really working on speech synthesis? How is it coming? Is it doable? Member from repair: yes we are. Speech synthesis is actually much more easier than speech recognition. Jason (from repair): It is more challenging to parse the SGML then to do speech synthesis. Jane: we have a person named Jeff that is an interactive designer who is an expert in speech synthesis who can help you (repair). Bob (client): we just want to make sure you do not underestimate the task that is posted in front of you in this speech synthesis issue. Jane: as I have said before, we have someone who has a lot of experience in this area. And we won't use the speech part if it turns out that it is not good enough.
Boundary Conditions
Speaker:Samuel Chong
No discussion(went out of time)
Action Items
For relevant teams: Add wearable computer and ACS to hardware/software requirements
For Tom Hawley: Get the IBM laptops and configure them like the real system we are going to be using. From this we can predict what software components we can squeeze into them exactly.
For Eric: to find out more about the speech synthesis
For Dan Heller: Revise System Design Document
Open Issues
Can Lotus Notes store wireframe?
How is persistent data stored in STARS system?
Is Speech Synthesis realistic?Are we underestimating its task?
Should we considered a pen-based system for inspection?
When are we going to get the GPS equipments?
Can IETM be downloaded into PEDD in small increments efficiently?
minute guy
Last modified: Wed Nov 17 03:53:35 EST 1999
This page is hosted by the Chair for Applied Software Engineering of the Technische Universität München. Imprint (Impressum)