I've got a budy. him and i have been playing with the idea of a RC AH-64 Apache. This would be flyed FPV with remote camera and headtracking goggles. this model would be made with fly by wire and this is where we lack the expertise that you have. would it be possible to make you balancing software transferable to flying a helicopter or/and a plane?
This robot uses light sensors and pid controlers. The light sensors won't be much use for you. Maybe the PID controlers will. Better sensors for a helicopter will be gyro and accelerometers (look also for my other robot NXT-body balance). But you better search for IMU and Helicopter on a search engine.
Thx, I started with manualy tuning(Lego Remote Control) while a supervised learner on the PC watched the process(through the IR-tower). The Learner(LWPR) could do the tuning quite well. No angles were calculated, only light sensor values.
The tilt is compensated with the wheel on the ground.
The roll is compensated with the one wheel in the air. The weight of this wheel changes the centre of gravity of the unicycle. But to move this wheel in the air you have to compensate for the moving force. There are the two turning wheels for. They compensate the force of moving the one wheel in the air.
I used the same sensor's as Steve Hassenplug(Hitechnic). And two RCX's to get a very small measure and control cycle.
Oh, I was way off then. I assumed you were using some type of incremental encoder sensors and I thought the wheel on the back was being used to sense roll but now I see the wheel in the back and the two spinning wheels banace each other's torsional forces.
Can you make this clearer? I've done some simulations with just the two turning wheel type of setup. I don't see what the third wheel in the back accomplishes; the two windmill-like wheels should be able to correct for roll by themselves.
BTW this is very impressive, so I'm not arguing with success! Just trying to understand.
Youre remark about the two windmill-like wheels will be able to correct for roll by themselves is correct. But the issue is that not the two windmill wheels correct the roll but the acceleration of this two. First the lego motors cant accellerate t0o much thats a problem because the cause of the roll is not solved. Cause of the roll are for example the instable rubber wheel or not a good offset from the sensors. So before the maximum acceleration is reached the centre of gravity is changed.
It seems the wheel in the back is being used to sense roll but what type and where is the pitch sensor? Doesn't look like it has any yaw control but do you plan on trying to get it to go anywhere? Not that balancing on a single wheel isn't cool enough but since you got the balancing working, it shouldn't be much more work getting it moving. Nice work though. I can only imagine how many hours of coding went into that :-)
This has been flagged as spam show
O Jesus, son of Mary! Is thy Lord able to send down for us a table spread with food from heaven
jesussonofmary1432.blogspot com/
ahmd1431 11 months ago
just so awsome :D
rocketman468 3 years ago
this is great but it seems to be wobbling alot, looks like it would fall over as soon as u touch it!
nattavision 3 years ago
awesome
Appelsappel 3 years ago
I've got a budy. him and i have been playing with the idea of a RC AH-64 Apache. This would be flyed FPV with remote camera and headtracking goggles. this model would be made with fly by wire and this is where we lack the expertise that you have. would it be possible to make you balancing software transferable to flying a helicopter or/and a plane?
Hope you still read this... : )
fedekaare 3 years ago
This robot uses light sensors and pid controlers. The light sensors won't be much use for you. Maybe the PID controlers will. Better sensors for a helicopter will be gyro and accelerometers (look also for my other robot NXT-body balance). But you better search for IMU and Helicopter on a search engine.
Good luck
Vegakade 3 years ago
i love it, especially since it is made form legos
RoboTekno 3 years ago
Amazing
Limpfittz 4 years ago
Absolutely a great robot!
How did you tune the PID controller parameters? Trial and error or some advanced technique?
Did you measure what the maximum recoverable angles of your model are? In both (tilt and roll) directions?
jvpommeren 4 years ago
Thx, I started with manualy tuning(Lego Remote Control) while a supervised learner on the PC watched the process(through the IR-tower). The Learner(LWPR) could do the tuning quite well. No angles were calculated, only light sensor values.
Vegakade 4 years ago
look on google for:
Locally Weighted Statistical Learning Software download
Vegakade 4 years ago
nice dude!
zedomax 5 years ago
pretty dam cool
unijamo5 5 years ago
please post the code!!
Also, what is the back wheel doing? What light sensors are you using?
davesum99 5 years ago
The tilt is compensated with the wheel on the ground.
The roll is compensated with the one wheel in the air. The weight of this wheel changes the centre of gravity of the unicycle. But to move this wheel in the air you have to compensate for the moving force. There are the two turning wheels for. They compensate the force of moving the one wheel in the air.
I used the same sensor's as Steve Hassenplug(Hitechnic). And two RCX's to get a very small measure and control cycle.
Vegakade 5 years ago
Oh, I was way off then. I assumed you were using some type of incremental encoder sensors and I thought the wheel on the back was being used to sense roll but now I see the wheel in the back and the two spinning wheels banace each other's torsional forces.
pintpi 5 years ago
Can you make this clearer? I've done some simulations with just the two turning wheel type of setup. I don't see what the third wheel in the back accomplishes; the two windmill-like wheels should be able to correct for roll by themselves.
BTW this is very impressive, so I'm not arguing with success! Just trying to understand.
davesum99 5 years ago
Youre remark about the two windmill-like wheels will be able to correct for roll by themselves is correct. But the issue is that not the two windmill wheels correct the roll but the acceleration of this two. First the lego motors cant accellerate t0o much thats a problem because the cause of the roll is not solved. Cause of the roll are for example the instable rubber wheel or not a good offset from the sensors. So before the maximum acceleration is reached the centre of gravity is changed.
Vegakade 5 years ago
Actualy the coding is quite simple. Just three PID-controlers in two RCX's. The construction was the challenge!
Vegakade 5 years ago
It seems the wheel in the back is being used to sense roll but what type and where is the pitch sensor? Doesn't look like it has any yaw control but do you plan on trying to get it to go anywhere? Not that balancing on a single wheel isn't cool enough but since you got the balancing working, it shouldn't be much more work getting it moving. Nice work though. I can only imagine how many hours of coding went into that :-)
pintpi 5 years ago
Is that balanced on a single wheel?
desiredusername 5 years ago
Yep, a single wheel
Vegakade 5 years ago
So the lateral balancing is from the arm? And this is all stock RIS (except for firmware)??
desiredusername 5 years ago