Interesting that the BEAM guy boldly declares that Integration (the I in PID) is not necessary for a balancing bot. From what I'm learning, Proportional control gives the biggest contribution to the balancing function, Derivative control mostly adds damping, and I believe that Integration controls overall drift from the starting position. So without Integration, the bot can balance, but will make no effort to return to its starting position.
Thanks Erco, your robotic ferreting skills are commendable. The Enicycle has one balancing wheel. That might be a great challenge for off-road endeavors but certainly exhilarating!
I wrote my first PID routine for the balancer and got it working with my homemade remote. Works quite well! I still have to improve my forward/reverse drive dynamics, but I have stationary balance & turns down pat now: http://www.youtube.com/watch?v=F7_JBBgsWrU
Cool video and thanks for the update. I have a few questions.
It looks like the batteries are at the top of the vertical? I'm guessing that's to increase the rotational inertia, so if an inverted pendulum is like a sailboat this increases stability. But it also seems like it would increase the required righting moment from your motors. How did you decide on the height?
It's interesting that the robot tends to creep counter clockwise. Is this because of motor differences and your PID algorithm is linear?
Yessir, batteries are on top to shift the CG higher. This is a pretty minimal bot, so the batteries represent a fair percentage of the weight. The high CG slows things down and makes balancing easier. Pretty standard fare for these bots.
I'm using quarter scale servos modified for CR. The good news: easy to use, built-in speed controllers. The bad news: 60 RPM max, and their center zero-speed position always creeps a bit, depending on sunspot activity and my checking account balance.
I can easily redefine the zero position in software. Works great for a while, then it starts creeping the other way . That's how I know someone cashed one of my checks...
Talk about gyro-stabilized! This guy's NXT Legobot spools up 4 lead-filled tires as a flywheel stabilizer, PLUS it has an optical active feedback balancing system: http://www.youtube.com/watch?v=KGqNeszVcdI
The NXT servo motors have quadrature encoders built-in, I'll bet that's how bots like Legway keep rotational errors at bay. They can likely also zero out positional errors as well. We just can't get away from wheel encoders.
The whirling dervish is riot, although inelegant. I've seen robot unicycles do that for side to side balance, but never a two wheeler.
A big cardboard standee on a balance bot would have lots of mass and air friction to act as a damper. Should slow things down and simplify things considerably. His students didn't finish their project, so I may have to jump in and try it. I have the perfect standee already. Stay tuned. I have to do SOMETHING for robotics week. Nothing else going on in LA!
That's pretty cool. It's much shakier than before. Do you think it is because the lack of rigidity in the board, or does the PID algorithm need to be re-tuned?
Phineas T. Ratchet is looking very good indeed! Sorry I missed this for a bit. Great job erco. Glad to know that Torrence is pulling its weight for National Robotics Week. You and P.J. came through in the nick of time.
Late night working on part 3 of the ROBOT magazine article on my Balance Bot. Finally got the driving software working pretty well; it's harder than it should be! http://www.youtube.com/watch?v=j-Ubi93VzHo
Looking good. I've liked following this project and all the intermediate steps. When you only see the final stage of a robot, you can fail to appreciate all the work that went into getting there.
@Ron: No sir, this one-sensor robot would have a tough time on an incline, unless you program in the exact angle and go perfectly straight up the hill. I'll leave that to the gyro/accelerometer Bots with Kalman filters that cost 20X what this sensor did!
I just want to get this bot driving smoothly on the flats. I'm a minimalist; I like to push a BS2 and this single sensor as far as I can.
A follow-up project in the back of mind is to modify a Scribbler (1) to balance. Mount the wheels & motors on stilts below the main platform and get it driveable just like this platform. That should breathe some new life into old Blue!
Comments
It even humps the guy's leg near the end!
Ken & Humanoido may find this interesting. Ramps, offroad, a neat vehicle that looks fun.
http://www.enicycle.com/
It looks like the batteries are at the top of the vertical? I'm guessing that's to increase the rotational inertia, so if an inverted pendulum is like a sailboat this increases stability. But it also seems like it would increase the required righting moment from your motors. How did you decide on the height?
It's interesting that the robot tends to creep counter clockwise. Is this because of motor differences and your PID algorithm is linear?
I'm using quarter scale servos modified for CR. The good news: easy to use, built-in speed controllers. The bad news: 60 RPM max, and their center zero-speed position always creeps a bit, depending on sunspot activity and my checking account balance.
I can easily redefine the zero position in software. Works great for a while, then it starts creeping the other way . That's how I know someone cashed one of my checks...
The whirling dervish is riot, although inelegant. I've seen robot unicycles do that for side to side balance, but never a two wheeler.
http://sports.yahoo.com/olympics/blog/fourth_place_medal/post/Leszek-Blanik-super-tumbler-and-internet-sensat?urn=oly-wp87
A big cardboard standee on a balance bot would have lots of mass and air friction to act as a damper. Should slow things down and simplify things considerably. His students didn't finish their project, so I may have to jump in and try it. I have the perfect standee already. Stay tuned. I have to do SOMETHING for robotics week. Nothing else going on in LA!
http://www.youtube.com/watch?v=Lr0MkFb3da4
Actually, I love that serendipitous flex; it's giving me all kinds of ideas for free animation from the sympathetic vibrations.
It is looking great! Are you going to try slight inclines?
I just want to get this bot driving smoothly on the flats. I'm a minimalist; I like to push a BS2 and this single sensor as far as I can.
A follow-up project in the back of mind is to modify a Scribbler (1) to balance. Mount the wheels & motors on stilts below the main platform and get it driveable just like this platform. That should breathe some new life into old Blue!