TAURUS2

[techy_pankaj]

hi!!

i went through the construction details of ur robo project taurus 2...there are some things i wud like to discuss...
i got a video in which the testing of the circuit is done..there i saw tht the wheels apart from rolling are also rotating bout the vertical axis...
so i was just wondering wat can made to do tht...i mean there has to be some constraints between all the 4 wheels so tht they take the robot in the single direction...but as all the four wheels were powered with different servos and they have their individual axis....wat can make them move so smoothly???
I am sorry if my question seems too dumb to you people...

and one more thing i want to ask is tht which code designer software and compiler and uploades i shud use in the strt...and which language i need to know for programming a bit tricky robots....

sorry for such lame questions but i wud be thankful if somebody cn help...its my strt to this field!!!

[gamefreak]

I figure for the wheels to turn vertically admin used 180 degree servos due to the fact that they have feedback control, he then used that feedback to control the direction of each wheel and tell the horizontal servos which direction to go as to not go against the rest of the bot.

Vertical= 180 degrees w/ feedback
horizontal= 360 degree servos

Also servos are pretty precise so i cant imagine that being to hard

EDIT:
(2) 7.2V nicad battery packs ($16)
(1) CMUcam ($100)
(1) Cerebellum Microcontroller, by Botrics ($40)
(3) homemade IR (~$4 in components)
(8 ) HS-311 servos ($71)
(1) null modem cable ($3)
(4) foam 4 inch diameter model aircraft wheels ($15)
(1) 1'x1' sheet of HDPE ($10)
(2) ball bearings (found in a scrap pile)
(1) 1'x1' sheet of aluminum ($10)

is the parts list, so im geussing that admin didnt expect for there to be high enough friction to offset a wheel

[Admin]

just to add on to gamefreak's comment . . .

the robot actually wasnt that hard to build - i only started making robots two years earlier than when I made T2. for the smooth line following operation I used a very simple fuzzy logic algorithm with a tweaked constant. i just needed to do that calculation for the left and right front wheels, then invert the angle for the rear wheels.

other motions, such as 360 degree turning and sideways straffing, i just hardcoded servo angles.

for the sinusoidal motion, i had the servo angles add +1 at each computational cycle until the angle maxed out, then had it subtract -1 at each computational cycle. i couldnt actually calculate sin(angle) because the PIC16F877 takes forever to do a single trig calculation (1/4th a second I think).

[gamefreak]

hey admin, what would you have done if something would of jammed the wheel from turning around the vertical axis.

You would have 1 wheel that wouldnt be hardcoded unless yo forced the servo back

[JesseWelling]

Another reason I like mechanical simplicity.

[Admin]

hey admin, what would you have done if something would of jammed the wheel from turning around the vertical axis.

You would have 1 wheel that wouldnt be hardcoded unless yo forced the servo back

well, servos are guarenteed to go to the proper angle, unless they break . . .

the bigger issue i had was my profuse use of superglue and packaging tape to build T2 added to a total lacking of apriori planning . . . i was such a bad engineer in those days . . .

[gamefreak]

servos move based on a pulse, or at least continous rotation do, With 180 degrees there is a feedback and i thought that you controlled them in a different way then just sending pulse after pulse to turn them.


if say you needed a continous rotation servo to go 90 degrees in 1 second and something held it in place for half a second then the servo would only get to 45 degrees.

So if you had continous rotation for the vertical axis then if something got it stuck momentarily then it would not be at what it should be.

right?

[Admin]

Oooohhhh ok I see what you meant.

Actually, my microcontroller never stopped sending commands to the servos, no matter what. So if something blocks the servo, as soon as the 'blockage' is removed, the servo will go to the correct angle.

The problem you mentioned would of course be a problem if at any point my microcontroller stopped sending signals to the servos during the time the servo was blocked.

[gamefreak]

i is confused, even if the microcontroller kept sending if the wheel got stuck at any point and you dont know what angle the wheel is at then it will start to lag behind the other wheels.


since the servos move based on the pulse duration and height or something liek that then:

send a pulse of 10 to all vertical servos
bottom right gets lagged for 5 of those so....
top right =10
top left=10
bottom right=5
bottm left=10

send pulse of -5
top right =5
top left=5
bottom right=0
bottm left=5


so if a continous rotation servo gets stuck then it will stay behind due to the fact that you have no way of knowing the angle of that servo.

[Admin]

Servos have internal feedback control. So if I tell it to go to 20 degrees, it wont stop rotating until its reached 20 degrees.

If its being commanded to go to 20 degrees and I try to force it with my hand to rotate, it will try its darnest to stay at 20 degrees.

Your microcontroller doesnt need to know what angle the servo is at, because it can trust the servo to do what it is told.

When you modify a servo, all you are really doing is tricking the servo feedback device that it isnt rotating when the servo really is. If I would tell a modified servo to go to 20 degrees, it will keep rotating thinking it hasnt reached the right angle yet.

[gamefreak]

see there it is, i thought you had continous rotation servos, all you have listed are standard servos.


so for the vertical axis do you have 180 degree rotation servos?

[Admin]

haha ok i see the confusion.

for the rotating wheels I use modified servos for continuous rotation. but for the wheel angles, I use non-modified 180 degree servos.

[gamefreak]

alright.

Now heres a question, why did you use a rocker design? it doesnt seem to have any usefulness to it.

[Admin]

three wheels are guaranteed to always be on a plane, but four wheels are not (its basic highschool geometry)

the rocker design makes sure all four wheels always remains on the ground even if the terrain is bumpy.

this is a problem with all vehicles with more than 3 wheels, such as cars (the car suspension is required to keep all four wheels on the ground)

my four wheel omni wheel robot (http://www.societyofrobots.com/robot_omni_wheel.shtml) also has a similar problem, and so i used a very thin aluminum base to act as a flexible suspension - if i didnt, only 3 wheels would ever contact the ground!

the rocker also gives a climbing ability to go over small obstacles.