Theo Jansen style robot leg: prototypes #4 and #5
January 1, 2009
This picture compares prototype #2 with an angled profile version:
The angled "L-profile" introduces a new property in the design: chirality. All of a sudden, it matters which direction you want to fold the joints. Compare the symmetric #2 on the right:
I decide to go with a "T-profile" and defer the hard choices by enlarging the surface area considerably. The scissors would later on decide which parts to "engineer away".
Here is one early version of the T profiled model. The large surface touching the table is intended to slide against the flat side of my robot later. The idea is to provide lateral stability. In other words: it will not fold its "knees" inwards or outwards despite the fact that it has only half the width of the boxed #2. At the same time, this design will benefit from joints that are hinging around a horizontal pin (paperclip wire in this case). Which I found to be the preferable pivot over the short pins in the "flat on flat" design in #3.
This is prototype #4 just before I close it up with the outer half of the shell. This half also has quite some surface area. This surface is designed to slide against the inner joints' profiles. All this sliding is going to cause a lot of friction 8-(
I pulled #4 apart again and traced its curvy shapes onto a fresh piece of cardboard. This time I inserted the "Holy Numbers" Theo Jansen published in his book "The Great Pretender" (click on image for larger version).
I had a lot of fun on "Old Year's Day" tweaking and assembling this version. The nice round curves add nothing to the functionality, but instill various mental images by association. The black tape was originally intended to prevent advertisement of the shop where I bought a lamp, but I decided it would also reduce friction, so I added some more.
Here's #5 almost fully assembled. It is the first model since #1 to get a crank shaft and accompanying links. And that false economy during the prototyping process cost me dearly.The bottom link overlaps with one of the struts inside the bottom joint. That "eye" in the copper wire is supposed to link with the pin hole down there.
This is where cardboard proves flexible and adjustable. I just opened up the strut. The next version probably will have that strut shortened and shifted. Here is the leg almost fully assembled:
And here it is with the crank shaft in place, clasping a pencil between its toes. I had to adjust the position of the shaft a lot or it would not be able to turn the entire revolution. The error is too big to blame just the use of inaccurate materials. The "holy numbers" had to be violated more than a bit! The radius of the crank was reduced as well.
I decided on the new postion of the shaft by "reverse engineering". I used a green pencil as a shaft and traced onto the box what the nearest and farthest positions could be. Then I chose an average position between the green scribbles as a reasonable place for the crank shaft.
With a shoe box as robot body, I placed the whole contraption against my bench and traced the revolution of the foot onto 5mm square paper (what else would I use?). The whole leg measures 260 mm from toe to upper pivot. This results in a stride that is 80 mm long and almost 18 mm high.
update 1 jan:
I cut some card board templates for transferring the pieces onto plywood (4mm).
Click on image for larger version of this "blueprint".
The numbers are dimensions of the link lengths in mm. Designations A-M are as pictured in Jansens book. His numbers are multiplied by 2. The designation "inside" is arbitrary and is relative to the vehicle: the "insides" of one leg all face towards the vehicle, or away.
The width of the struts is also a bit arbitrary. Total leg width shall be 32 mm in this case. The 24 mm struts are 32 - 4 - 4, where 4 equals the plywood thickness. The other struts are another 4 mm narrower = 20 mm.
Note: these templates are yet untested.
update 3 jan:
Well, I started testing the templates. And they failed. On a few minor details. The leg is not completely assembled yet. Tomorrow I will cut out the final pieces and glue them together. Here are more pictures.
Traced the templates onto 1/7 inch (3.6 mm) plywood.
And cut them out with my new fav tool: a table jig saw. I only wished I had more than one blade for this. I am still looking for a supplier for those (it's a FERM FFZ 400R, if you happen to work at a hardware store here in Zuid Holland).
This is the parts partly assembled. Compared to the templates. I like wood a lot better than cardboard. It's just prettier!
Notice the screw up here? It's that damn chirality again! The template shows how I should have glued them together. The plywood one is obviously "inside out". Back to my fav tool 8-(
Below follow two pictures of the almost fully assembled leg. I am waiting until I get it all right before I start gluing that double decker part again! I added two plywood cranks. Wood just feels better than copper wire. But it is also chunkier and I need to allow for the wider cranks to move around. This means adjusting the position and size of three "struts" (is that even the correct word?). Good thing I am not using any superduper glue.
Also, note the black marks in the middle where I need to take away a bit more to allow the leg to fold all the way to the left.The strut that should be in the double decker needs some serious cutting up to allow for that crank "K". I am rethinking the curve in that one. A straight piece would allow for a much longer strut, which I need for the parts to touch each other and to provide stiffness in the leg. That would mean that strut "I" must be shorter. The top crank "J" needs a sharper curve on the inside.
Most extreme leftward position pictured (should go further left):
Most extreme rightward position (OK):
Now, what would I use for a crank shaft / axle / hub / drive / system ?
All pages tagged with "theo jansen" here.