mini segway project

[Rickertsen2]

Now that the build season is over i am looking for some interesting project to occupy my time. One thing that came to mind was a mini rc segway about a foot or so tall. I figure it will make a fun toy and challenging MISO control problem.

In order for an inverted pendulum to remain stable, the sum of the horizontal forces must equal zero. On a segway, the rider shifts their mass forward to signal that they want to go forward. This moves the horizontal center of mass of the system in front of the point where the wheels touch the ground. If the segway did not accelerate, the rider and the segway would fall forward. Acceleration leads to a counterforce that balances out the falling while at the same time increasing the velocity of the system. I would assumme that eventually the segway reaches a maximun velocity and stops accelerating at which point the center of mass must once again be above the wheels. The decelleration process is the reverse.

All of this is just inference and assumptions. I have never even ridden a segway before and don't really have a clue how the balancing works.

This is all fine and dandy until you remove the rider as would be the case in my robot. Now you have nothing to shift the center of mass in front of the the wheels during acceleration. The only way i can think of to make the thing drive forward is to first jerk backward. I cannot fathom a reasonably intuitive remote control scheme that would accomplish this. I have considered an actuated counterweight but would prefer to avoid this.

http://www.segway.com/products/rmp/
That works somehow without a counterweight

[ahecht]

You might want to look at the Legway code, since that incorporates the ability to do things such as line tracking (which implies riderless movement). He does it exactly the way you describe, by moving backwards slightly to offset the inverted pendulum.

[sanddrag]

1. Ride a Segway
2. Be amazed
3. Say,"I can do that for around $1000"
4. Try to build your own.
5. Quickly discover why the real Segway costs so much
6. Abandon the project and save your money.
7. Buy the real thing.

I have completed step 5.

[ZZII 527]

This sounds like a really fun project! Keep in mind that keeping a mini Segway balanced is actually more difficult than the real thing. Think about trying to balance a pencil on the tip of you finger versus a broomstick. But it is by no means impossible.

As for getting it started: When it's normally balanced and stationary, the drive motors are constantly canceling out minor displacements forward and backward. If you suddenly ALLOW a small amount of forward displacement (while still restraining backward tilting) it would naturally fall forward a bit. It would provide enough torque at the wheels to balance this offset and start moving forward. When you are done accelerating, take out the offset and it should return to constant velocity. Easier said than done...but it might be a starting point.

[David55]

This page should help you:

http://www.geology.smu.edu/~dpa-www/robo/nbot/


And here is some info from the segway site:

http://www.segway.com/segway/how_it_works.html
http://www.segway.com/segway/component_details.html

[EricRobodox]

You got to post pics when you are done. I think a segway that a pet can ride would be pretty funny. lol. You may want to look into gyroscopes that will fit the project. Since it will be a lot smaller, a foot you said, than try and make sure that the motors, speedcontrollers, gyroscopes, battery, and everything else can fit.

[amos229]

i would try finding an rc parachutist and using the servos in tthat control your wieght issue. another thing u could consoder is using an rc gyro also, however i really dont know much about custom electronics so ggood luck.

[Tom Bottiglieri]

Has anyone ever thought about using a servo neat the top of the balancing bot to shift the platform's center of mass? This would probably simplify things code wise, because all you would need is a handler to change the position of the servo, and the robot to always try and balance itself.. instead of the robot trying to balance and modifying the wheel speed to drive which way you want.

[irishninja]

You could also use ultrasonic sensors, so when it tips forward the ultrasonic sensor is at an angle and can "see" the ground. You know if the ultrasonic sensor is on the front or back, you can accelerate in either direction based on this knowledge.

[EricRobodox]

problem there, going up hills, like a slight hill. Wont work that well without some major fine tunning, programming, and engineering. I dont recomend the ultrasonic sensor idea.

[MattK]

This sounds like a fun project and all the power to ya! Just please keep in mind any and all safety hazards if you or a friend decides to "ride" it (I know its a "mini segway" right now)! Sometimes experimental balancing machines can go a little funky, and its really no fun to be moving around 10mph and have the machine fail- sending you pounding down into the pavement at that speed.