lagache@violet.berkeley.edu (Edouard Lagache) (04/17/88)
I attended a lecture by Hubert Dreyfus on the problems in
Artificial Intelligence, and he mentioned that he was aware of only 2
Expert systems that work as well or better than the human experts that
they were based on. What does this have to do with trains? Well, one
of the systems (called ALPS) is designed to optimally load a cargo
planes, which is a problem that looks isomorphic with the problem of
loading a railroad switch yard.
That raises an interesting question for those interested in
computers and trains: what sort of expert systems have developed for
the railroad industry? It seems to me that there are a number of
promising areas:
1.) Scheduling.
2.) Optimal switching moves and train assembly.
3.) Cargo routing and loading.
4.) Equipment Maintenance.
Does anyone know of what work (if any) has been done by railroads
or A.I. outfits in this area? Interestingly enough, Dreyfus would
probably claim that the first 3 areas would be very promising domains
for expert systems.
Edouard Lagache
School of Education
U.C. Berkeley
lagache@violet.berkeley.edu
P.S. I has posted this to both 'rec.railroad', and 'comp.ai'. Please
don't reply to both groups unless it is truly of general interest.jaw@mtgzy.UUCP (XMRN60000[bsm]-j.a.welsh) (04/19/88)
> What sort of expert systems have developed for the railroad > industry? Strangely enough, the one that I know of is a General Electric locomotive maintenance expert system. It was mentioned in a computer magazine and one of the railfanning mags. last year.
tony_mak_makonnen@cup.portal.com (04/20/88)
> What sort of expert systems have developed for the railroad > industry? Strangely enough, the one that I know of is a General Electric locomotive maintenance expert system. It was mentioned in a computer magazine and one of the railfanning mags. last year. yes and it was finally coded in Forth.
welty@steinmetz.ge.com (richard welty) (04/21/88)
Someone wrote: * What sort of expert systems have developed for the railroad * industry? In article <4643@cup.portal.com> tony_mak_makonnen@cup.portal.com writes: >Strangely enough, the one that I know of is a General Electric locomotive >maintenance expert system. It was mentioned in a computer magazine and >one of the railfanning mags. last year. >yes and it was finally coded in Forth. It was written here at GE R&D, a few years back, for use on GE locomotives (in a group that was a precursor to the AI program that I work in now. I was not involved in the project itself, which was completed long before I joined the program). Reportedly, it has had good success in speeding up diagnosis of problems in locomotives in the field. -- Richard Welty Phone H: 518-237-6307 W: 518-387-6346 welty@ge-crd.ARPA {rochester,philabs,uunet}!steinmetz!welty ``Avalanche is better than none'' -- Bullwinkle Moose
ceb@edai.ed.ac.uk (Colin Bridgewater) (04/22/88)
In article <8816@agate.BERKELEY.EDU> lagache@violet.berkeley.edu (Edouard Lagache) writes: > > ....for those interested in > computers and trains: what sort of expert systems have developed for > the railroad industry? It seems to me that there are a number of > promising areas: > > 1.) Scheduling. > > 2.) Optimal switching moves and train assembly. > > 3.) Cargo routing and loading. > > 4.) Equipment Maintenance. > > Does anyone know of what work (if any) has been done by railroads > or A.I. outfits in this area? Interestingly enough, Dreyfus would > probably claim that the first 3 areas would be very promising domains > for expert systems. Just to get my two penn'orth in, whatever happened to dynamic programming for scheduling, cargo-space optimisation and inventory control etc ? This well-worn technique is quite adequate for the majority of purposes envisaged by EL. I mention this to raise a wider issue which was possibly not in the mind of the original sender, namely that of the desire to throw ever more complex solution procedures at the simplest of problems.... Why should we want to implement an expert system, when adequate techniques exist already ? That is, is the application of expert system technology appropriate to the magnitude and complexity of the problem ? Should we be advocating the application of such 'high-tech' solutions to all and sundry ? I have no doubt that such systems could be made to work, don't get me wrong on that, I just question whether the level of technology required in order to do so is justified. Surely it is better to apply the simplest solutions when- ever possible. Having said that, I too, would be interested to hear of any research, actual implementations etc that are around. As an engineer involved in AI, I look for simple solutions, in the (vain ?) hope of being able to debug them when things go wrong.......... Colin Bridgewater Univ of Edinburgh P.S. there is an expert system around that diagnoses faults and discusses repair strategies on diesel-electric locomotives. Unfortunately, I don't have any references to hand, but I hope that this jogs someones memory. The Happy Hacker loves to go a-wandering, it's legal in the UK (official).
ed298-ak@violet.berkeley.edu (Edouard Lagache) (04/26/88)
In article <73@edai.ed.ac.uk> ceb@edai (Colin Bridgewater) writes: > Just to get my two penn'orth in, whatever happened to dynamic programming >for scheduling, cargo-space optimisation and inventory control etc ? This >well-worn technique is quite adequate for the majority of purposes envisaged >by EL. I mention this to raise a wider issue which was possibly not in the >mind of the original sender, namely that of the desire to throw ever more >complex solution procedures at the simplest of problems.... > > Why should we want to implement an expert system, when adequate techniques >exist already ? That is, is the application of expert system technology >appropriate to the magnitude and complexity of the problem ? Should we be >advocating the application of such 'high-tech' solutions to all and sundry ? >I have no doubt that such systems could be made to work, don't get me wrong >on that, I just question whether the level of technology required in order to >do so is justified. Surely it is better to apply the simplest solutions when- >ever possible. > Dr. Bridgewater comments are not completely off the mark. One reason I posted the question was to see if Expert System methodologies might be useful in improving the performance of conventional programming techniques by providing useful heuristics for such tasks as switching. While, the areas mentions can clearly be solved by brute force methods, it is unlikely that human experts employ only those sorts of stategies (since human cognition doesn't support large active data structures); thus, there may be some interesting enhancements possible on conventional programming techniques by learning how human experts perform the tasks involved. At the same time it is very much in keeping with Hubert Dreyfus's comments that just railroad tasks are very promising areas for expert systems that will outperform human experts. Any comments? Edouard Lagache The PROLOG Forum lagache@violet.berkeley.edu
smoliar@vaxa.isi.edu (Stephen Smoliar) (04/26/88)
In article <73@edai.ed.ac.uk> ceb@edai (Colin Bridgewater) writes: >In article <8816@agate.BERKELEY.EDU> lagache@violet.berkeley.edu (Edouard >Lagache) writes: >> >> > ....for those interested in >> computers and trains: what sort of expert systems have developed for >> the railroad industry? It seems to me that there are a number of >> promising areas: >> >> 1.) Scheduling. >> >> 2.) Optimal switching moves and train assembly. >> >> 3.) Cargo routing and loading. >> >> 4.) Equipment Maintenance. >> >> Does anyone know of what work (if any) has been done by railroads >> or A.I. outfits in this area? Interestingly enough, Dreyfus would >> probably claim that the first 3 areas would be very promising domains >> for expert systems. > > > Just to get my two penn'orth in, whatever happened to dynamic programming >for scheduling, cargo-space optimisation and inventory control etc ? This >well-worn technique is quite adequate for the majority of purposes envisaged >by EL. I mention this to raise a wider issue which was possibly not in the >mind of the original sender, namely that of the desire to throw ever more >complex solution procedures at the simplest of problems.... > > Why should we want to implement an expert system, when adequate techniques >exist already ? That is, is the application of expert system technology >appropriate to the magnitude and complexity of the problem ? Should we be >advocating the application of such 'high-tech' solutions to all and sundry ? >I have no doubt that such systems could be made to work, don't get me wrong >on that, I just question whether the level of technology required in order to >do so is justified. Surely it is better to apply the simplest solutions when- >ever possible. > There is one issue of "appropriate technology" which appears to have been overlooked in Colin's argument; and that is the matter of computational tractability. In many practical domains, while it is certainly possible to build mathematical models which may then be processed by dynamic programming, those models are too unwieldy to yield much useful information in any reasonable period of time. Often what makes an expert an expert is the ability to recognize that a complex general-purpose model may be considerably simplified through abstraction without significantly sacrificing fidelity. The mathematical nature of the model, in and of itself, cannot provide us with information of how to perform such abstractions. That is often why we need experts; and, in such cases, if that expertise can be properly modeled by an expert system, a computationally intractable approach can be turned into a practical one.
gilbert@cs.glasgow.ac.uk (Gilbert Cockton) (04/27/88)
In article <9226@agate.BERKELEY.EDU> lagache@violet.berkeley.edu.UUCP (Edouard Lagache) writes: >> Why should we want to implement an expert system, when adequate techniques >>exist already ? > While, the areas mentions can clearly be solved by brute force methods, > it is unlikely that human experts employ only those sorts of stategies > (since human cognition doesn't support large active data structures); > thus, there may be some interesting enhancements possible on > conventional programming techniques by learning how human experts > perform the tasks involved. a) Expert systems use large active data structures. b) Knowledge representation has not yet captured human problem solving methods - at best they model them at the black box level. c) We do not know how people solve problems, we only have observations about behaviour under certain (often fake lab.) conditions. d) If a mathematically sound technique is slow but sound, it is usually preferable to heuristic-based faster but less accurate solutions. With some AI tuning techniques, no measure of the technique's limits can be given, whereas in classic Operational Research, the trade-offs of quick and dirty methods can be expressed quantitatively. I know of a problem where a mathematics department said there was no solution to the problem, so some people started to look and see if there was an AI solution instead - i.e. a rule-based approach. Does anyone know of why such an approach is credible and worth pursuing? Surely when mathematicians say a problem has no known solution, it's rather naive to muck about with a mathematical object like a computer to see if it gets you anywhere? In the railway case, what grounds are there for thinking that 4 knowledge engineers and an expert system shell are going to outperform a decade of OR research?
roberts@cognos.uucp (Robert Stanley) (04/27/88)
In article <4643@cup.portal.com> tony_mak_makonnen@cup.portal.com writes: >> What sort of expert systems have developed for the railroad >> industry? > >Strangely enough, the one that I know of is a General Electric locomotive >maintenance expert system. It was mentioned in a computer magazine and >one of the railfanning mags. last year. >yes and it was finally coded in Forth. I missed the original post because our feed has been down for a while, so can't e-mail a reply. The system referred to was D.E.L.T.A. (Diesel- Electric Locomotive Troubleshooting Assistant), and not only was it re-coded in FORTH, it was delivered on a pretty basic IBM-PC. An interesting feature was that the system as installed in G.E.'s workshops included a full set of locomotive schematics stored on optical disk, and the user was pointed to the correct area by D.E.L.T.A. The system was not developed in FORTH, but once the knowledge base had been completed (they don't introduce new designs of diesel-electric locos very often) it became feasible to build a "conventional" re-implementation. FORTH was chosen for various technical reasons, including applicability of the stack approach, and excellent run-time performance. The development was done in LISP on something like a PDP-11/23. The original reference was: Bonissone and Johnson "Expert System for Diesel Electric Locomotive Repair" - IJCAI-83 I know that CN (Canadian National) have a small but active expert system group, who have produced several small systems. One was a diagnostic system for walkie-talkies (they use tens of thousands), and another was some kind of locomotive fuel usage monitor/advisor. I haven't been in touch with the group recently, so am not up on current work. CAIP Co (Canadian Artificial Intelligence Products Corp) has a joint agreement with CP Rail (Canadian Pacific) for the marketing of a lube oil expert system. The Transportation Development Centre (TDC) of Transport Canada in Montreal is active in the expert system field, but I am not sure whether they have any projects specifically for the rail sector. The Japanese have been extremely active in this area, but I'd need to look out a bibliography from a couple of years back for direct references. A number of interesting systems have been developed in France, which has an advanced railway system. The same is true in Britain, but I would have to do some paper file searching for details. The only system which springs to mind is British Telecomm's amazing British-Rail timetable advisor, which used speech recognition and voice synthesis for unrestricted access via telephone. If the original poster wants more detailed information, please would (s)he contact me via e-mail. My apologies for cluttering the net. Robert_S -- Robert Stanley - Cognos Incorporated: P.O. Box 9707, 3755 Riverside Drive Compuserve: 76174,3024 Ottawa, Ontario K1G 3Z4, CANADA uucp: decvax!utzoo!dciem!nrcaer!cognos!roberts Voice: (613)738-1440(Research) arpa/internet: roberts%cognos.uucp@uunet.uu.net FAX: (613)738-0002
Barry_A_Stevens@cup.portal.com (05/24/88)
A man with Canadian Pacific Railway in Montreal has built an expert system
which works in conjunction with a gas chromatograph and/or spectrograph to
perform an analysis of the oil in large engines. By looking for the metals
that are carried in the oil, expensive engine failures are found in
advance. They are claiming some impressive savings.
The details are not at hand now. If interested, please call or write,
I'll try to help.
Barry Stevens
Applied AI Systems
PO Box 2747
Del Mar CA 92014
619-755-7231
{_