I was just finishing up a report on fuzzy logic when a co-worker handed me a copy of your article "What's All This Fuzzy Logic Stuff, Anyhow?" in the May 13 issue. In short, I agree. It is extremely refreshing to hear a voice of reason in this age of emperor's new clothing.
I have included the first version of said report. It covers the development of a fuzzy-logic controller (and that of a corresponding classical controller) for the balancing of an inverted pendulum attached to the end of a robotic arm. The effort was requested by NASA in order to aid in the evaluation of fuzzy-logic-control-system design techniques. I hope you can find time to read it (at least the results and conclusions).
In the article, the use of the term "deterministic" to describe conventional controllers is unfortunate because it implies that fuzzy-logic controllers are not deterministic. In fact, fuzzy-logic controllers are completely deterministic. More specifically, the end result of a fuzzy-logic control system (assuming no add-ons or hybrids) is simply a function. That is to say, it generates a unique value of any output variable (e.g. current to a motor) for every unique set of values on the input variables (e.g. angular position and rate of the motor shaft as feedback variables).
Fuzzy logic is simply a design technique that is based on the use of intuition in designing a control function—a method to convert fuzzy human statements about how to control something into a completely deterministic relationship between controller inputs and outputs. Now, any control function can be developed in many different ways (a linear function can even be easily modified to be nonlinear if desired, what with today's modern computers!), and the performance is only dependent on the function, not the technique used to develop it. Thus the real issue is whether one technique or another is more helpful to the designer in developing a function. What technique(s) is used is primarily a matter of taste for the designer. Statements such as "fuzzy-logic controllers perform better than conventional controllers" are nonsense.
Finally, let me say that the idea of using fuzzy logic to convert human intuitive statements into a working function initially caught my fancy (for various reasons). However, this is far more complicated than it sounds. In my development of two very different control systems (one being a highly nonlinear "phase plane" controller for thruster-actuated space vehicles) with fuzzy logic and other control techniques, fuzzy logic proved to be a very poor technique. It took me far longer to develop fuzzy-logic systems, that achieved the same performance (and almost identical resulting functions) as with the other techniques that I used. The designer must mentally juggle many numbers and rules in specifying the fuzzy-logic system, and understanding the relationships between these and the observed system behaviors can be very frustrating indeed.
Lawrence E. Merkel, Lockheed-ESC, Houston, Texas
Isn't it funny—the F.L. approach is always very easy for F.L. guys who don't know how to do it any other way! I suspect that making F.L. work well isn't as simple as they say. Thanks for your observations.—RAP
I recently read an article that talked about the lack of "feel" for engineering in the current crop of engineering students. It disturbed me that this condition should exist. The article goes on to lament that many of the students had never dissected nor constructed anything before graduation, and that special courses were being designed for this purpose. What in the world are they doing in the field if they have gotten as far as college and have not dissected anything or constructed anything?
When I was nine years old, I constructed my "soap box" racer out of an old crate, a 2-by-4, and a roller skate that I had taken apart. I even had a headlight that I had made from the lens/bulb assembly of an old flashlight and a large type "A" cell. I did all this without any input from adults. I copied the design from "soap box" racers that I had seen kids use in Brooklyn, N.Y.
In 1955, when I was in the 8th grade, I got an "Honorable Mention" for my science fair exhibit, "The Transistor, an Electronic Marvel." I constructed a two-transistor audio amplifier (using Raytheon CK722s) and explained its operation and the operation of the transistor. I suppose it really was an engineering project and not scientific research. I did it on my own initiative and without any adult guidance. Maybe I would have done better if I had some guidance, but maybe it would not have served me well in life.
As a fifteen year old, I built a shortwave converter for my AM radio. I got the circuit schematic from an amateur radio handbook. I purchased the components, did the sheet metal work, and assembled it by myself. There was no adult involvement here either. I still have it and it still works.
I don't know if you are familiar with the famous "Canal Street" in New York City. At least it was famous in the '50s for its surplus electronics outlets. I used to make many pilgrimages there in my youth. On one trip, a friend and I carried home a 400-Hz generator "safari style," i.e., slung from a pole we carried between us on our shoulders. He needed it for his surplus ham equipment. We carried it this way on the Staten Island Ferry and then onto the New Jersey Central Railroad for the ride home. I had picked up a bargain mercury thermometer, which promptly got broken when I got caught in a swinging door at the ferry terminal. Ouch, a whole buck down the tubes....
The examples I could cite are almost endless. You may say that my experience was unique. Well, at the time it wasn't. Most of the people I knew in the College of Engineering at Rutgers University were either hams or hobbyists or experimenters. I, and others in my lab group, could not believe it when one of our lab partners asked us what a resistor looked like. He had done all the theoretical calculations in the various classes, but had never actually handled any of the parts. He was unique, though, and caught on quickly. He later got his PhD in Physics and was doing heat balance work on communications satellites.
You may say that kids today don't do that kind of thing anymore. Well, you would be wrong. My son, who's a junior in high school, worked on robotics projects at home during the summer. He has an old VIC-20 computer and a Fischer-Techniks kit. He is building working models of machine tools and other gadgets. He has assembled a siren, an AM radio, a burglar alarm, model rockets, and other goodies from kits. He has taken "junk" electronics and made them functional. He has dissected countless mechanical gadgets. He works on his bicycle by himself. He has the "soul" of an engineer. I think it comes from within. I don't think you can train it into someone.
So, tell me what is really wrong with today's crop of engineers? Is it the age of instant gratification, where building a kit model airplane takes too long? Is it the lack of role models? After all, anyone today who likes math and science is considered a nerd. In my day, I was classified as an "egg-head." But I got along with everyone: jocks, hoods, and the "in" crowd. I did not consider it a "put down." How would I fare today?
Finally, I am appalled that the article states, "Employers visiting the campus stress that it is more their communication skills that help students advance than engineering know how." I suppose you can design junk, but as long as you present it well, it's OK. I wonder if you value communications skills over engineering skills?
Why am I rambling on like this? Well, I feel that you are a kindred soul. You, and others like you, seem to be a breed that is slowly dying out. Any comments?
Alexander F. Jaszek, Needham, Mass.
Well, since most electronic stuff is "non-repairable," and there are no more Heathkits, it's hard to get kids interested in electronics—unless you count computers. There are still some bright kids coming along. Maybe we don't need more engineers, but we could sure use some smart ones!—RAP