circular motion

[ahmedragia21]

Hi, im trying to make a curve using one wheel in the forward and tune it to get an angle and use two motors on one axis , but i've failed , i made constant angel , sometimes the robot do the right curve till it hits a tower and many times to go to another position , also i noticed that the positioning of the sensor in the start area affects , is there a solution to this or change the whole system to use two motors on two wheel seprated and do PWM on both wheels ?

thanks

 

[Sceadwian]

Controlling two wheels independantly will help, especially if they're stepper motors, regular motors are going to require some sort of feedback to control the angle they turn to reliably, that's why many robot makers use steppers as drive wheels. They're not very fast, but they can be controlled very precisley.

 

[ahmedragia21]

So my original design is incorrect and wont help ?

 

[dknguyen]

I don't understand what you mean by two motors on one axis. Are you talking about tricycle-type steering?

 

[ahmedragia21]

Sorry if i didnt clarify
that's what i mean
| o |
| |
o--x---o
where o are two wheels and one front wheel which can be tuned to make an angel to follow a curve
x is one motor

 

[dknguyen]

Sorry if i didnt clarify
that's what i mean
| o |
| |
o--x---o
where o are two wheels and one front wheel which can be tuned to make an angel to follow a curve
x is one motor

A robot with two wheels, and one motor for each wheel will be very agile. But without feedback from each motor it can be very difficult to drive in a straight line (even with feedback it won't be perfect because of small differences between the wheels, motors, and gearboxes).

A tricycle robot lets you use one big powerful motor to move the robot forward, and another "steering motor" (that must have absolute position feedback) will steer a third wheel (that is not powered) that controls the direction of the robot. It is less agile, but if you have an accurate steering motor it will let you drive in a straight line better. But tricycle robots can tip over more easily, especially when you are moving fast and/or turning at sharp angles. The advantage is that you only need feedback on one motor/wheel (the steering motor/wheel).

So there is probably nothing wrong with your design, you probably just need a more accurate steering motor or sensor. There's nothing wrong with your design. It just probably needs to be built more precisely. I personally prefer a very agile two-wheeled robot over a tricycle robot. Why don't you just use an RC servo motor for the steering motor? They're made for that, and quite accurate too.

A tricycle robot WILL make a circular path more consistently than a two-wheeled robot. (THough the circle may be of the wrong size).

Robot Builder's Bonanza: Over 100 ... - Google Book Search

Can you post a sketch of your sensor setup?please).

 

[ahmedragia21]

i dont use any sensors , and i dont use any motor in the front wheel , i just tune it by my hands and stick it then test , some times it did 6 expeirments correct !! and another time it didnt any correct curve .. i may change the whole system today and use differential drive system then do PWM on both wheels , would this get me better results ?

 

[dknguyen]

So you are only using one motor on your robot? And you are just bolting the front wheel in place at an angle to have it travel in a curve?

WIth no feedback, a 3-wheel robot will make a more consistent turn than a 2-wheel robot. But if you have no motor on the front wheel to steer your 3-wheel robot, a two-wheel robot with two motors and two PWM at least can make turns at different angles rather than just travelling in a blind circle all the time so it would be more flexible than the 3-wheel robot. But without sensors it would be worse at following the curve to hit the wall than the 3-wheel robot without sensors that you have right now.

Ask yourself this, could you blindly walk around with your eyes closed following a curve to hit a tower? Probably not very consistently (and you know where your feet are!). Feet don't slip and wheels do. So your robot has to be able to "know" where it's wheels are to have any hope of doing it as well as you do. Or you could skip that step and give it some eyes of some kind.

If you had two front wheels and bolted them at an angle (like a car with a locked steering wheel) you might have better luck following the curve more consistently. REmember- the inside wheel has to be at a higher angle than the outside wheel since it travels on a smaller circular path. You'll have to do trial and error (or some geometry) to get both front wheels to be the right combination to turn the robot on the same path.

 

[ahmedragia21]

thanks for reply but what kind of feedback , speed feedback ? or how do i determine how should i put the robot in the exact position because i've found that any sligh angle error when putting the robot in the start point , it do a wrong curve , i think this problem will happen in all soultions

 

[dknguyen]

thanks for reply but what kind of feedback , speed feedback ? or how do i determine how should i put the robot in the exact position because i've found that any sligh angle error when putting the robot in the start point , it do a wrong curve , i think this problem will happen in all soultions

Not speed feedback as much as position feedback of the motors (though either will probably work). Even then, nothing will substitute the robot actually being able to *see* where it's going and adapt to unexpected changes in it's path.

THe best thing would be something that can measure it's wheel rotations to approximate the path it has taken so far, and "eyes" to see the tower. THat way it can sort of follow the path you want it to follow AND actually reach where you want it to go.

What are you trying to do anywys?

 

[ahmedragia21]

Im trying to do a curve to hit a tower.
Well , so its just position feedback ? but how do i able to see the tower ? using ultrasonic or what ? two of it ?

 

[dknguyen]

Im trying to do a curve to hit a tower.
Well , so its just position feedback ? but how do i able to see the tower ? using ultrasonic or what ? two of it ?

I do not know. Most of the reliable methods involve making the tower different somehow so the robot knows it is the tower like with an IR beacon or something. Just "seeing" the tower with something like a camera is quite a bit harder.

 

[ahmedragia21]

anyway i will try the pwm method today and i will tell you the news , and thanks for your great help