2 wheeled balancing robot

[John Gutmann]

i am planning on making a 2 wheeled balancing robot and was wondering if any body could help me out in terms of the mathematical portion becaus e i am VERY good at math but i am only entering math b (one year b4 pre calc) and I dont know what formulas i need

i am planning on using some sort of gyro to balance

[Adam Richards]

Would the robot be ridden on, autonomous, or just a standard controlled robot?

[mechanicalbrain]

I thought you were doing a walking robot? Also list you size. We need details, details and more details. What size would be a good start (the variables you need to use changes with size.) Small sizes are much easier than large. Very small is the easiest of all (yes that is not ALWAYS the case but mostly true).

[Ryan Foley]

Thats quite the project. Just so happens that another CD member is/was working on the same type of thing. There's a thread that he created, that probably has the answers you're looking for, or at least some useful insight that may lead you to the answers you need. Check it out here

Good Luck!

[Mike]

Originally Posted by Ryan Foley

Thats quite the project. Just so happens that another CD member is/was working on the same type of thing. There's a thread that he created, that probably has the answers you're looking for, or at least some useful insight that may lead you to the answers you need. Check it out here

Good Luck!

Haha, thats me!

When I was doing research, it looks like there will be a decent amount of noise from the gyro. Many people recommended using a kalman filter, which looks pretty complex.

You may want to try using something like IR distance sensors, the Sharp is good I heard.

My idea's been moved to my ever-growing to do list, so I probably won't get it done for a while. If you need any help feel free to contact me, or when you're done be sure to tell me how it worked out

[John Gutmann]

Originally Posted by mechanicalbrain

I thought you were doing a walking robot? Also list you size. We need details, details and more details. What size would be a good start (the variables you need to use changes with size.) Small sizes are much easier than large. Very small is the easiest of all (yes that is not ALWAYS the case but mostly true).

i asked someone and they said that a this would be easier for me then a bipod

plus i like the mth involved in this, hahaha first sign of a geek!

also are there any enigneers or people that can help me on aim msn or yahoo with the math, i know the outline of what i need to do (linearize the the inpout voltage and the tilt) but i have figured out how yet

[BrianBSL]

Originally Posted by sparksandtabs

i asked someone and they said that a this would be easier for me then a bipod

plus i like the mth involved in this, hahaha first sign of a geek!

also are there any enigneers or people that can help me on aim msn or yahoo with the math, i know the outline of what i need to do (linearize the the inpout voltage and the tilt) but i have figured out how yet

First, we need to know much more about what you are doing this with - what size, what controller, what sensor(s), what motors.

There is absolutely no equation someone can give you that is the motor output as a function of the analog input that will instantly work, epically without knowing that information.

As I told you when you IM'd me, I think you are thinking of this way too much like a scientist and not like an engineer. I would not be concerned with some equation that backs up the physics behind it, but rather something that provides a solution - which are related but not the same. You will probably need some PID control, or some sort of error-based control theory, and watch your integration term so you don't get oscillation. Without Calculus, I think you will have difficulty understanding the math of the physics that explains the problem, and I think you might want to take a simpler approach. I don't understand what you are talking about when you say "linearize the input voltage and the tilt" - you are getting discreetly sampled values on the analog input. The tilt angle is the integration of the yaw rate sensor's values, in whatever units they are in (you will have to convert the 10-bit analog values to degrees or something it if you want to use a non-arbitrary unit). You probably want to break the problem up - find a way to keep the robot at a stand still without falling, and then worry about driving it without it falling, having it move at the same time is an additional task to worry about after you can get it to stand still.

Others, please correct me if I am wrong, I have not actually done this project, but this is how I would approach it.

[John Gutmann]

well it is not that i want an equation to back up the physics but i am doing what you said in the IM. You said for me to do the opposite of the input. I get an input as a degree measurement, which means i need some way to convert it to a force that can be turned in to a voltage then applied. The math is for the conversion.

[Mike]

Originally Posted by sparksandtabs

well it is not that i want an equasion to back up the physics but i am doing what you said in the IM, you said for me to do the opposite of the input, i get an input as a degree measurement, which means i need some way to convert it to a force that can be turned in to a voltage then applied, the math is for the conversion

I think spellcheck just exploded.

I would help you, if I had any idea what you are saying.

[John Gutmann]

Originally Posted by Mike

I think spellcheck just exploded.

haha, what is spelled wrong?

[K.Shaw]

Originally Posted by sparksandtabs

haha, what is spelled wrong?

equation not equasion

[henryBsick]

Back on topic... Mike said try using IR distance sensors... good idea. Another may be trying to go as uber simple as you can. Get a shaft encoder and gear is down (for better resolution to the encoder once the shaft rotates) to a shaft with which a counter balance would hang. The robot tips, the "pendulum" stays perpendicular, rotating the shaft, spinning the gears, in which case the encoder would then detect rotation. Just code it out like a joystick from there on, with a lot of tweaking to get it just right. The robot tips one way, the encoder reads it and send power to the wheels to keep it upright. It will probably be a very wobly robot though, without a lot of resolution on the encoder to detect the smallest change, also there must be a large "power curve" to the motors, to make very small adjustments at first, then get larger as the rotation of the shaft becomes greater.

Sorry if that cam out in a jumble, I have been thinking about it for a long time now. They are just my personal non-programmer, Lego oriented views.

P.S. I really really really think this is possible with Lego's and am going to hop on it as an Independent Study Project as soon as school starts again.


Anxious for School,
-Henry

[John Gutmann]

Originally Posted by K.Shaw

equation not equasion

haha that is what i get for doing it late at night

[John Gutmann]

i am looking to do it with small gear motors or servos, MAYBE i am not sure how much power i will need so i may use CIM or drill motors and i dont know what i will use for a controller because i dont know the speed yet, the size will be about 2 feet tall or less and it will be a robot that will not bbe riden on i will eventually make it so i can control it like adding an offest to set the "balance" a little off so it will drive one way while it is "balanced"

would i be able to use folmulas from the inverted pendulum problem? because there motrs would be moving it and the gyro would meaure the tilt just as a pot on a inverted pendulum would

[John Wanninger]

Here's one way to approach it (or at least get started):

You are trying to "drive under" the center of mass. If you had a pendulum connected to the shaft of a potentiometer attached to the robot, it would give a certain voltage when you are directly underneath it, lets say 2.0 Volts. If the robot is tilted to the "North" 10 degrees, you get 1.8 Volts, to the "South" 10 degrees you get 2.2 Volts. If your robot tilts South, drive South , if it tilts North drive North. Try setting your motor command in proportion to the difference between the pot. signal and your pot.target signal. This example shows Kfactor as 1, adjust as needed. This is basic proportional control, the "P" in PID. If you want your robot to move South, adjust your target to something greater than 2 Volts so that your robot is "falling" in the direction you want it to travel.

To add integral control ("I"), add a constant whenever it is falling too far in one direction, subtract it if it is falling in the other direction. To add derivative control ("D"), subtract off a term proportional to the rate of change of the pot.signal [IMG]Diagram[/IMG]