Hi everybody.

As part of a graduation project, in which we are making a go kart drive itself and navigate an unknown terrain, I have to install dc motors on the gas and brake pedals to control the kart's speed. I have given it a lot of thought and I think that the best control approach is to use fuzzy logic control.

For example:
If (speed is more than required and is getting larger) Then (press the brake pedal)

I haven't used fuzzy logic control before but I read about it and I feel it will be right for this application because it will be very difficult for me to mathematically model the system in order to 'classically' control it. Please tell me what you think and whether this approach could yield good results or not.

Thank you very much.

 

I heard fuzzy logic has lost steam in recent years. In my opinion, it is not going to be any easier no matter what approach that you take. The best way is to keep it simple as possible.

For practical time to completion, go with what you know, or get someone with knowledge to help you. It takes a lot of microprocessing power to have a robot navigate through terrain. People have tried and spent years doing research on the subject, so that you know what you are up against here. I would recommend downsizing your project to a manageable level like navigate around a maze with walls or start with a robot like the one Sony has developed. Good luck--sorry can't be much help.

OJ

 

I kinda disagree. Only if people keep looking will we find the elegant, simple solution that will have everyone saying "Oh yeah, why didn't I think of that!" The examples are legion.

Regards
cww

 

Hello [email protected],
There are many examples, but Fuzzy is not the case.
The name of ship defines its lot. ;-)

--
Best regards.
Vladimir E. Zyubin
Sunday, September 11, 2005, 7:18:25 PM

 

A friend of mine once said, "Fuzzy Logic applies approximate solutions to little or misunderstood problems."

I did a paper on Fuzzy Logic and its ability to compete in the Motion Control world... It couldn't.

Just as an FYI, there is a Government prize of $2,000,000 for anyone, any place, any company, etc., that can develop a completely autonomous vehicle which can traverse unknown terrain. The contest is known as DARPA. You can find it using your search engine.

What you are about to embark on is not trivial. I agree with OJ, that since this is a school project in which time and money are the issue, he has indicated the optimum approach.

I also agree with cww, however, that giving up too soon is not necessarily a good thing. It could have delayed the onset of technology as we know it for years... we might still be riding horses!

ChuckR . . . www.EngSolu.com

 

FL is just one of statistic methods of modelling. As well as Neural Networks, and others... so called "data mine" technologies.

There was much ado about it. But it seems to me the main application it can be used was (is and will) control of washing machines... that can be easily implemented by classical methods, BTW.

Have no plan to offend anybody.

Best regards.

 

Thank you for your comments.

I am very aware of how hard this is going to be and about the previously-done researches in this field. There will be an on-board computer taking care of the heavy computations. I was just wondering about the best control approach for controlling the speed through pressing the brake and gas pedals. If I don't use fuzzy logic, how will I control it otherwise or even attempt to model the whole thing?

Thanks.

Nichola Victor Abdo

 

I did not suggest giving up any where.
I merely being practical by saying that for a student project, a robot that can navigate through rough terrain is way too difficult and time consuming [years to finish].

There is an article by Bob Pease from National Semiconductor on Fuzzy Logic. If I recalled correctly, he dismissed it because no one could prove that it is something other than the classical control system with rules applied to it.

OJ

 

Ok let me put it this way. Forget about navigation and rough terrains. Just assume the problem is pressing the gas and brake pedals of a car in order to maintain a desired speed. Would fuzzy logic be suitable for such a thing alone? I DO NOT intend to use fuzzy logic to navigate or control the steering of the car; just as a subsystem for speed control.

Nichola V. Abdo

 

Initially I was a sceptic and thought fuzzy logic was a new age 'crock', until a colleague of mine started looking into Fuzzy Logic and showed me an example he had developed on a heating test rig. He had a standard PID controller in a CompactLogix for comparison and his Fuzzy controls in a structured text file. It has turned my ideas about fuzzy logic.

Some sites and references he sent to me are:
Fuzzy Application Library/Technical Applications/Practical Design Practical Fuzzy Logic Design by Constantin von Altrock (8/96)

Citation Reference: This paper was published at the Embedded Systems Conference in 1996 in Paris. The Fuzzy Logic Application Note series is published by Inform Software Corporation on its Internet server to promote the use of fuzzy logic technologies in applications.

http://www.fuzzytech.com/e/e_a_pfd.html

Fuzzy Shower Demo
http://www.iit.nrc.ca/IR_public/fuzzy/fuzzyShower.html

Fuzzy Pendulum Demo
http://www.iit.nrc.ca/IR_public/fuzzy/FuzzyPendulum.html

Crane load sway demo
http://www.clarkson.edu/~esazonov/neural_fuzzy/loadsway/LoadSway.htm
There are rules that can be changed on the load sway demo.

http://www.industry.siemens.de/water/en/solutions/sector_fuzzy-logic.htm

I still impressed with the terms 'Fuzzy-fication'

I would think your application is suited to this type of control. We are looking at implementing it on the next project with a difficult type of process.

Cheers,
Trevor Ousey

 

It's an interesting problem because you have two different actuators and depending on the rate of decelleration you will need to use the brake or decrease the gas pedal. I admit I don't know much about fuzzy logic. Every explanation I've read is in fact too fuzzy.

You might just close the gas loop as in a traditional manner but if the commanded decrease in speed is not being met in time (conditional integral) you can close the brake pedal loop and disactivate the gas pedal loop. You may also want to have another conditional that if the velocity error is too great that you want to activate the brake loop immediately (This is of course only true if the commanded velocity is less than the actual velocity in your loop, i.e. you need to decellerate fast).

Who knows, maybe what I'm describing is fuzzy logic and I just don't know it, but it seems more like conditional control algorithms to me.

~Ken

 

In reply to Trevor Ousey, your comment reminded me that one of the applications for fuzzy logic is heater controls.

 

In reply to Nichola V. Abdo, one of the early demonstrations of fuzzy logic was in acceleration and braking control of a passenger train in Japan. You might wish to research references on that application.

 

In reply to Nichola V. Abdo, one of the early demonstrations of fuzzy logic was in acceleration and braking control of a passenger train in Japan. You might wish to research references on that application.

 

hai,

With a minimum knowledge of fuzzy control my suggestion to speed control using fuzzy logic is given below.

You can define 2 major inputs for the fuzzy controller --> e= speed error, de = present acceleration/decceleration. Now you have to define different quantisation levels (fuzzification levels) for the same. Using a suitable
membership function & a defuzzification algorithm (llr to center of gravity method) generate different fuzzy levels for the output.Now this output implimentation (usually change of output) depends on two quantities -- close of the gas loop & close the brake pedal. You can use logic to generate the actions on both the quantities depending on the fuzzy o/p.

You can combine the fuzzy controller with neural networks to obtain optimised performance for the system.

With Regards,

Fahid K.A.
Sr.Project Engineer,
Kalki Communication Technologies Pvt Ltd,
#147, Anathapushpa Building,
5th Main Road, 7th Sector,HSR Layout,
Bangalore 560 034, India.

Phone: +91-80-2572 1263/64
Email: [email protected]
Fax: +91-80-2572 5473
Web: www.kalkitech.com

 

"I did a paper on Fuzzy Logic and its ability to compete in the Motion Control world... It couldn't."

Can you elaborate?

-Will Dwinnell
http://will.dwinnell.com

 

"There is an article by Bob Pease from National Semiconductor on Fuzzy Logic. If I recalled correctly, he dismissed it because no one could prove that it is something other than the classical control system with rules applied to it."

...and programming languages are "nothing more" than abstractions of machine leanguage. Does this mean they are not useful?

-Will Dwinnell
http://will.dwinnell.com

 

'FL is just one of statistic methods of modelling. As well as Neural Networks, and others... so called "data mine" technologies.

There was much ado about it. But it seems to me the main application it can be used was (is and will) control of washing machines... '

Fuzzy logic is currently used in a wide variety of applications. I'm not sure what you mean about washing machines, but another poster has already included references to motion control applications and there are wide array of examples of fuzzy logic being put to good use in non-control applications (image processing, GIS, forecasting, etc.).

-Will Dwinnell
http://will.dwinnell.com

 

On September 25, 2005, Will Dwinnell wrote:
> I'm not sure what you mean about washing machines, <

http://www.google.ru/search?hl=ru&q=Fuzzy+Logic+Washing&btnG=Поиск+в+Google&lr=

> applications and there are wide array of examples of fuzzy logic being
> put to good use in non-control applications (image processing, GIS, forecasting, etc.). <

There is the only problem... FL is not suit for the control tasks. You can do it, as wall as program in machine code, or assembler... Human beings think in terms of the human logic, not fuzzy or Mars ones.

--
Best regards.
= Vladimir E. Zyubin
= Monday, September 26, 2005, 1:45:14 PM =

 

> FL is not suit for the control tasks. <

Can you explain why you think so?

-Will Dwinnell
http://will.dwinnell.com