I really had no choice in the platform. We used the 18F452 (and supporting development board shown in video) because it was a code development board that was under consideration for a digital electronics lab for the Physics program.
The most robust fuzzy logic controllers I have seen are computer controlled. We ran into speed and memory restrictions while developing but at the time I was quite new at C coding.
The DC motor, h-bridge and pot were all purchased from digikey. The dev board and bread board were supplied by the instructor and the rest was manufactured in the machine shop.
We were originally going to use an optical encoder but instead chose to use a single turn 10K potentiometer. I added a tune-able voltage divider to be able to calibrate the top-dead-center at 2.5V.
This project was more of a proof of concept and because of time constraints we did not do a complete analysis of the system. It was our understanding that one of the strengths of a fuzzy logic controlled system is the ability to control without model specific analysis.
One major problem is that I didnt have time make use of azimuthal measurement. If, for example, the arm was rotating left under inertia and the rod was balanced perfectly, then the rules dictated that the arm should stop. This causes the arm to fall over due to momentum and a corrective action begins. Its a major problem but I didnt have time to correct it.
NOW would be a great time to ask for federal bucks to further fuzzy logic study because everyone in Washington is using it!
mazakman1957 3 years ago
I really had no choice in the platform. We used the 18F452 (and supporting development board shown in video) because it was a code development board that was under consideration for a digital electronics lab for the Physics program.
The most robust fuzzy logic controllers I have seen are computer controlled. We ran into speed and memory restrictions while developing but at the time I was quite new at C coding.
mgrunert 3 years ago
The DC motor, h-bridge and pot were all purchased from digikey. The dev board and bread board were supplied by the instructor and the rest was manufactured in the machine shop.
mgrunert 3 years ago
thanks a lot
I m planning to implement a fuzzy logic based inverted pendulum
what do u suggest wud be better
a PC or shud we use a microcontroller
and where did u get the physical components
can u elaborate on that plz?
urmish 3 years ago
We were originally going to use an optical encoder but instead chose to use a single turn 10K potentiometer. I added a tune-able voltage divider to be able to calibrate the top-dead-center at 2.5V.
This project was more of a proof of concept and because of time constraints we did not do a complete analysis of the system. It was our understanding that one of the strengths of a fuzzy logic controlled system is the ability to control without model specific analysis.
mgrunert 3 years ago
how do u convert angle into digital data?
and did u study the dynamics(i.e differential equations etc) ?
urmish 3 years ago
What's the source of oscillation in this system? Backlash, long flexible pendulum with low frequency mode, quantization error?
uniment 3 years ago
One major problem is that I didnt have time make use of azimuthal measurement. If, for example, the arm was rotating left under inertia and the rod was balanced perfectly, then the rules dictated that the arm should stop. This causes the arm to fall over due to momentum and a corrective action begins. Its a major problem but I didnt have time to correct it.
mgrunert 3 years ago
Ah, that explains it. Cool stuff. I'm in the process of making my own, hopefully I'll get it working soon =)
uniment 3 years ago