Let's Make Robots!

"Moooo" Rare Swiss Cow Project has begun

Following on from the Laser Experiments i have decided to attach it to some kind of walking Robot.

I am now a avid fan of Theo Jansen -  after rik brought him to light in his Blog.

So far i have Prototyped two pairs of "Theo" legs using "MindStorms" as an experimental base.

It worked better than i could have imagined - these legs go like a Cow on its way to the milking parlour.........

Only problem now is what  medium shall i use to make the final version.....

.....and how the blasers will i get it to turn - may have to run it off 4 motors and double up on the number of legs.

First stage is to replace the Lego Mindstoms with something more exciting..... Wood.

 

This worked well ..... however after looking closely it suddenly dawned on me

that there are two fixed triangles in the arrangement .....dduuhhhh,

( it puzzled me to rik why you coloured those bits in green )

 



 

 

 

 

so i simplified the mechanism with well yes 2 triangles........and 2 flexible squares.

I used the Hypothenus as the fixed link static anchor points (yellow motor and wood spar)

 

 

 

 

The motor is a geared motor courtesy of "OddBot"

Using some PolyMorph i was able to quickly construct a crank arm,

with PolyMorph it is so easy to tweak the orientation and length, 

giving a super smooth gliding effect....

Second Crank arm made and the yellow motors changed to GM2s for lower speed and higher torque.

Also at long last it has four Legs - and the arduino with motor controller driving them.

 

 

 


 

 

 

 Made 2 extra legs and for more torque installed two GM2 motors

 

 

 

 

 Syncronising the Legs :-

I had a problem with the missmatch of the legs - ie she would not walk straight with equal power going to each motor.

So i installed 2 QRD1114s, one pointing at a flat region with painted line on each motor drive shaft.

So both motors send a sync pulse every revolution - if these occur at the same time then all is well and the leg motion keeps a straight line.

When there is a difference between the arrival of the two sync signals a control action is made - stop one motor (slave) and wait for other motor (master) to catch up- then restart slave motor - back into sync and cruising straight again.

 

I have a number of other strategies - all based off the QRD1114s which will unfold.........any other ideas greatly welcome.

 Time for some cosmetic surgery - Swiss Cow gets Her "Beauty Treatment"

Using first a Cardboard Template

i crafted the PM with some "Skinning" techniques.

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.
Gareth's picture

 Time to forge forward and Head in the Right Direction...........

...... so what i thought would be a rough Plan "A"  "see how it looked" head .............

Moohead__2_.jpg

...turned out better than i had thought...... a Calf is born...........

Moohead__10_.jpg

From the rough calf template i was able to create a stable PM framework....

.... with a movable jawbone (dont ask why ..... it seemed a good idea at the time).. 

Moohead__13_.jpg

 

 Just as back up i made a "Paper mache" as a plan "C".....its still drying on the radiator i think....

Moohead__15_.jpg

 

 Using PM base - i "skinned" the gaps and fattened the cheeks.......

i used the cardboard ears as templates for the PM ... the feeling of the ears are exatly like "Grissle"

The Eyes....yes the eyes are very Special.....

.. the first one who guesses where they come from will get a free pint from thier local Irish pub.......

Moohead__17_.jpgMoohead__18_.jpg

Great job! I guess the eyes are taken from deodorant bottle (roll)?

Synchronization of the legs looks great, how well it works in turning ?

Gareth's picture

Deodorant bootle glass/plastic thingy it is not..... though from now on i am saving them.....thanks (they call out to have leds installed - maybe a marketing idea there)

The "Thynk" control is tuned only for straight line work, for turning this has to be blocked as turning is acheived by stopping one set of legs whislt driving the other (its a form of skid steer).

I am still installing the Payloads at moment so until i know how heavy she is on her hooves - i have no idea of the final turning ability. (for sure it will turn on flat shiny surfaces ..... carpets or grass is anybodys guess !!!!)

Gareth's picture

I was bit worried about the lack of "straightline" ability - because of a combo of maybe motor missmatch or leg frictions.

mo_005.jpgSo i came up with a simple Sync Control using QRD1114s.

Basic idea :-

  1. Each motor shaft has a reflective light sensor
  2. one pulse off each motor for every rotation
  3. When sync pulses arrive together all is well (legs working as they should)
  4. When sync pulses arrive at different times all is not well and a corrective action is made
  5. Corrective action :-
  6. Stop Slave motor
  7. wait untill Master motor catches back up
  8. then restart Slave moter - meaning Legs are back in sync.


This works fine (in fact better than calculated) ..........

........ next idea is measure the millis of each rotation of both master and slave and tweak the speed of one to bring back into sync

Gareth's picture

This is not so easy to explain ....... i try this way

  1. With crank arms vertically down legs are 200mm wide apart. (four legs touch ground)
  2. With crank arms vertically up legs are 70mm close together. (four legs touch ground)
  3. This alone is pretty amazing as the crank arm is only 25mm.
  4. The 70mm set up allows a stable contact to floor at all times.(ie his minimum standing base is 70mmx120mm)
  5. Whenever the crank arms are "in-sync" (any position) all four legs are on the floor.
  6. Whenever the crank arms are "out of sync" only two legs are truely in contact with floor (because he cannot balance on two legs he tips a little to compensate so in reality three legs contact the floor, forth one waves in the air).
  7. Forward locomotion seems to be a case of weight transfer and forward dragging of the front leading leg.
adding rubber "hooves" puts her under stress - it needs to Skid-"Steer" (hehe) a little.
OddBot's picture

Hmm... Can't quite remember the name of the fellow, famous dude known for working in mysterious ways. Not related by any chance?  :p

I would like to know if it still turns with rubber feet to grip the floor.

Gareth's picture

At momment rubber Hooves are confusing the leg movement (or confusing my idea of the true leg movement).

..... i am leaning into the idea this beast needs to skid steer a bit.

However my plan it to attach some longer legs onto the triangle knee bits ..... maybe this weekend.......

g_code's picture
I have been watching him for a couple years now, amazing work.
OddBot's picture
I loved this video, it was great! ( I am bias as your videos are about the only ones I see these days). I think you might need to make a spider cow (8 legs) to get good movement without falling over. Well done!
Gareth's picture

I had also calculated on 8 legs but i am happy to say it works well on just 4 (saves a bit of work i can say),

Have also added new Video of it going forward and also turning (though its a mystery how it works :-)

This is current strategy for turning :-

For it to go straight it need the crank arms to be 180° out of sync........

For it to turn left or right  the phase has to change from 180° to 90° (90° is max speed turn - on the spot)