Let's Make Robots!

Gakken interview - for the record

Byrik

Dear Rik,

Thank you again for cooperating in our magazine.
(And also for introducing about us in your blog too!)

Following is the questionarie that I want you to ask about your Theo Jansen style robot leg.

Dear Miki,

It's my pleasure to help you with your articles. I will send you my
answers in different e-mails. That way you have your text as soon as I
write it. If I would save it all up, you might have to wait too long for
the entire text.

--
Question 1
When and how did you first know about Theo Jansen and his "creatures"?
What did you feel when you first saw them?
And why did you decide to create Theo Jansen style robot leg?
(In your blog, you wrote that you were looking for an alternative locomotion
 for your symbiotic robots, what other locomotion did you tried for the robot?
 And why did you choose plywood as a material?)

When I saw Theo Jansen speaking on ted.com [1], I was mesmerized by the
motion of the dozens of legs on a single machine. To me it looked like a
swarm of flamingos as seen through a tele lens. It is impossible to
follow the path of a single leg. And yet, the design is based on the
principle of an individual leg. This must have been in the summer of
2008. As I discovered later, I had been introduced to Theo Jansen's
writing many years before that. He used to write a column in a Dutch
newspaper [2], which I enjoyed a lot. He introduced me to "genetic
algorithms", which are an under appreciated aspect of his strandbeest works.

The impulse from the Ted talk came at the same time as I discovered the
online community letsmakerobots.com. I have always been fascinated by
clever, original, unconventional designs. So I tried to adopt Jansen's
design for many legs and use the single-leg-design for something I could
consider "designed by me with a little bit of help from the coolest
designers in the world". I have a similar fascination for the locomotion
methods created by Robert Full c.s. [3]. His "Rhex" demonstrates why
hexapods are evolutionary unavoidable solutions. A walking hexapod
always has the bare minimum of three legs on the ground at any given
moment. The tripod is a perfectly stable platform. A hexapod is
basically two tripods alternating. (Note the difference between engineer
Jansen and engineer Full: artist and teacher.)

My very first video on Youtube (ever) simply demonstrates a new kind of
leg to the robot makers of LetsMakeRobots [8]. I got hooked and I tried
different things, designs, materials. My video skills also improved
along the way. This explains how I got to study the strandbeest
mechanics. As that happened, I felt the desire to actually use the
mechanics for a proper robot. And although that never fully happened, I
am very glad that I spent so much of my hobby time on this project. It's
been very rewarding, I suppose because it appeals to many different
senses at once: my sense of logic, geometry, mechnics, my sense of
beauty, my sense of making things with my own hands, my sense for
sharing ideas with a community. The sharing part of it has been more
important than I realized at first. My online community (LetsMakeRobots)
provides me with much of the energy I put into my projects. And in turn,
my blog has inspired a few others to build their versions of a
strandbeest. Notice that not a single creation looks the same!

I am still envisioning a vehicle (or creature) with six or more
"TJ-legs", three on each side. It will have a cargo space between the
leg: a fundamental difference from any strandbeest I've seen so far.
Perhaps a fundamental flaw on my part. I want to find out for myself and
then share my findings with the world. I estimate that my personal
teacher:artist ratio is approximately 4:1. :-)

I never looked for other means of locomotion for my robot. I started out
with the locomotion and I decided that I would build a robot around it
some time. Somewhere deep down my fantasies, I dream of two robots
living in symbiosis. One bot will a tractor/rover: very mobile and
agile, with powerful motors and lots of sensors. But it will run out of
power fast. The other robot will be a trailer, almost completely
stationary. It will have some kind of power plant on board. Probably a
wind turbine. The large heavy batteries on board the second robot will
power the big motors of the first robot. When the two bots couple [4],
they will travel together to a place where the wind is more powerful.
Once there, the bots will separate again. With the rover scouting for
more wind while the wind turbine recharges the communal batteries. Those
heavy batteries could sit between two rows of TJ-legs, but never on top
of them. The fact that Jansen designed his legs specifically to avoid
lifting the body vertically makes them very appealing. That is a great
energy saver for any heavy cargo carrying vehicle. [5]

Plywood? Oh well, it was a logical step after cardboard. I like to
design parts in 2D and then assemble them into 3D shapes. It leaves the
possibility to create them with CNC machines, home made or otherwise.
[6] By using it so intensively, I learned a lot about its good and bad
properties. Also it looks beautiful and accepts all kinds of finishing:
paint, wax, oil. I probably would have chosen something different had I
not owned a table jig saw. My inventions are designed for at least 25%
by random events like that. Go with what you've got. Let your personal
imagination and inventiveness fill in the blanks. That's what I like so
much about Jamie Mantzel's attitude. [7] It also appears to be one of
Theo Jansen's defining character trades.

------

Oh boy! Did I just write all of that!?!!?? :-)
Good luck editing this down to an article! Feel free to be VERY
selective. It's fun writing this. I think I will post this on my blog
some day. I promise to wait until you published your magazine though.

[1] http://www.ted.com/talks/theo_jansen_creates_new_creatures.html
[2] http://opinie.volkskrant.nl/author/show/id/99/Theo_Jansen
[3] http://www.ted.com/index.php/talks/robert_full_on_engineering_and_evolution.html
[4] http://letsmakerobots.com/node/2315
[5] http://www.google.com/images?q=Animaris+Rhinoceros
[6] http://www.thingiverse.com/tool:31
[7] http://www.youtube.com/user/JMEMantzel
[8] http://letsmakerobots.com/node/2480

Question 2
Please tell us about the mechanism of your Theo Jansen style robot leg.
You've inserted Theo Jansen's "Holy Numbers" in prototype #4,
did they help the moves of robot leg's to improve?

When I started to use plywood, I knew that my new material would not be
as forgiving as cardboard. Cardboard is a great prototyping and
designing material because it bends and folds. The folds in cardboard
can be used as hinges and pivots. No extra parts required! But in
plywood I needed to build pivots and I needed all the geometric
proportions to be just right. Jansen's "Holy Numbers" were already
proven to be "just right". So when I found them online, I stored them
somewhere safe.
Those numbers are not very easy to come by. It's not that Jansen is
keeping them top secret, but he is also not giving them away like free
candy either. Out of respect for that attitude, I buried my version of
the geometry in my own website. They are findable, but not easily so.
Hopefully the search for the numbers will make people realise that they
are curating something very magical.

I numbered my subsequent prototypes 1 through 10. Every new model has
some improvements over the previous one. But the geometry of the
triangles between the pivot points always remained true to Jansen's
original design. I do not dare to mess them up. I learned that a little
bit of compliance is still required to make the mechanism turn a full
revolution. At some positions it is almost impossible to turn the crank
further. All the different leverages inside the leg are then working
against the motor's force. I tried to remedy that by tweaking my
implementation (sanding, lubrication, even banging with a hammer).
Reducing friction is still an important task for my leg. The choice for
plywood resulted in friction problems. Deal with it Rik!

Also scaling down the size of the leg (mine is 28 cm tall) has greatly
affected the leverage of each link (bone) in the leg. Jansen's legs are
taller than he is, but the pivots are almost the same diameter as mine.
His bones turn around 19 mm pivots, mine are 10 mm. His are PVC plastic,
mine are plywood and hardwood. I may have made a design mistake there!

Question 3
What are the peculiar features of your robot leg compared with
other Theo Jansen inspired works that other creator's had made?

Almost every Theo Jansen Strandbeest inspired work (exception here [1])
that I've seen, arranges the legs like Jansen did. Lots of them, side by
side, all driven by a central crank shaft. The only payload that such a
machine could ever carry, would have to sit on top. (Some people say
machine, others say creature. I use the words as synonyms in this context.)

My design aims to have plenty of room between the legs. They each have
their own crank. Three or more legs could be mounted flush against the
side of a box shapes structure. Like my cardboard prototype was mounted
against a shoe box. The idea is that a flat leg pressed against a flat
surface will still have sufficient lateral stability. And that a payload
between the legs (motor, battery, electronics, sensors, perhaps even a
manipulator or a baby robot) will lower the centre of gravity, thus
improving stability.

I am paying dearly for this radical redesign: friction, complex
mechanics, more parts. Every improvement I made in the last 3 or 4
design iterations, were aimed to reduce the complexity and parts count.
I changed from brass to wooden pivots for example. And I changed the
plywood parts so that they do not require screws to stay together. It's
literally a jig saw puzzle. The only remaining metal parts make up the
power transmission system from motor to legs. Maybe if I scaled up my
legs by a factor of two or more, I could use plywood gears for the
transmission.

The arrangement of legs on a creature can be varied without limit. We
have brainstormed about it on LetsMakeRobots a lot. [2], [3] However
remaining property of the TJ-leg is always the intrinsic synchronisation
of the legs' movement. The legs stay in phase with one another. This
makes it so beautiful and functional at the same time.

[1] http://letsmakerobots.com/node/13709
[2] http://letsmakerobots.com/node/6725
[3] http://letsmakerobots.com/node/6101

Question 4
After introducing your work on our magazine, we guess some of our readers
will try making the legs using plywood like you did.
What kind of advice would you give to your follower?

Advice on plywood...
Buy the good stuff. You will be working the material a lot. Six legs
times twelve parts makes for a minimum of 72 items to saw. You will want
the plywood to be easy to saw. Buy one small sheet and test it. Or
several sheets of different kinds and test them all. I now have five
different sorts of plywood in my workshop. And I learned a lot about the
kinds that my local shops sell me. I cannot possibly judge the plywood
available on the other end of the world. So here are some general notes.
Thickness: Thicker plywood makes it easier to drill holes exactly
perpendicular. This is important for pivots.
Wood variety: hardwood might look prettier, but it is not necessarily
sturdier than softwood varieties. Plus hardwood is tougher on your jig
saw (or CNC router or laser). For example, I discovered that my hardwood
ply was bonded with a different adhesive. After 30 minutes of sawing I
had inhaled so much of that smell that I started to develop a head ache!
Ventilate your workshop well! It turned out that density of the ply is
more important than the hardness. A finer wood grain gives greater
accuracy in the small details like pivot holes.
Build up three dimensional shapes by stacking differently shaped layers
of ply. Glue them together. Use tiny holes and nails (1 mm) to position
them with great precision. Remove those nails after gluing.
Create a set of "golden templates" and copy the working parts off of
those templates. Include the "positioning holes" in your templates. Use
the hardwood for templates. That will make them durable and accurate.
Your readers are encouraged to copy my designs. They are published in my
blog [1].
They are also encouraged to share the results with other enthusiasts
online. English speaking makers are specifically invited to do so on
that very website. My most important lesson during this whole journey
was the value of support from friendly, like minded "fools". :-)
And that's true for any material you choose.

[1] http://letsmakerobots.com/node/604

Question 5
The last question, what are the points that Theo Jansen's work impresses you?
Do they serve as a stimulus for your creation?

The allure of Theo Jansen's Strandbeest is multi faceted. I was first
attracted to the beautiful choreography of the many, many legs. Then
fascinated by the geometry of the individual leg. Then I learned about
the genetic algorithm that produced the "optimum" ratios of all the
dimensions in the leg. I started to understand what the *engineer*
Jansen tried to achieve. Why a mere set of two "right isosceles
triangles" and one "equilateral parallelogram" will not suffice. Why
indeed the number are "Holy". I started to appreciate the ideal of not
ever lifting your weight while making a step. And finally, I learned the
hard way how difficult it can be to reproduce something so deceivingly
simple.

Being a small part of an Internet phenomenon made me realise how many
special people out there are making the effort to make something like a
strandbeest. I feel a great appreciation for all their efforts and
results. A simple search on youtube for "strandbeest" or "theo jansen"
will keep me entertained for hours on end. It would greatly inspire me
(again) to see many new additions to the genus "animaris" out there.

Aside from all the engineering marvel I gained a little insight in the
*artist* Jansen. Or at least into his art. The philosophical aspects of
his work are certainly not lost on me. He calls his strandbeesten
"creatures" and he is only in part trying to provoke his audience. Or at
least, that's how I interpret his art and his explanation of it. There's
a fascinating juxtaposition in his work. The engineering can be
analysed, improved, replicated. But the art is undefinable, even
debatable. Just as art ought to be.

Compare music. Any physicist can explain sound waves to you. But nobody
can explain the way it changes your mood. Theo Jansen is like a Hans
Dulfer constantly inventing new saxophones. And then playing them.
Beautifully.

Thank you again for your cooperation!

Best regards,
Miki Hayashi

This hopefully answered all your questions, so far. Feel free to ask for
clarifications or additional images (if your deadline permits it). It's
been a joy to revisit this 18 month old project.

Finally, could I be so bold as to request one copy of your magazine. I'm
sure it cannot be obtained from my local periodicals kiosk.

Regards,
Rik


 


 

All of that got reduced to this:

Which was translated for me by the ever friendly and polite Mr Miki Hayashi!

 


TITLE
Theo Jansen Walker

SPECIFICATION INFO.
Producer : rik
Material : Plywood, Hardwood
Size : 280mm full length
URL : http://letsmakerobots.com/blog/1069

MAIN TEXT
"Jansen designed his legs specifically to avoid lifting the body vertically makes his beests looks appealing. But that is a great energy saver for any cargo carrying vehicle." says rik, who is thinking if he could develop the Jansen's link to practical use. He envisioned of a vehicle with legs on each side and a cargo space between the leg. After trying many trials, this leg with artistic figure was born -- but it is not completed yet. Though rik had forgotten about this project for 18 months, he is still envisoning the vehicle with six or more "TJ-legs". An "evolved" strandbeest carrying cargos --- it's hard to imagine its looks, but must be a creature loaded with many dreams.

CAPTION
Pictures of his successive trials, movies introducing the moves of the leg, and even a picture showing the track by putting a LED to the tip of the leg (photo taken from http://letsmakerobots.com/node/7164 ), and even the
In his blog, you can see great many images and movies of his successive trials, such as the plan of the leg's part is shown on his blog. It's worth visiting to see the beautiful moves that his legs makes.

 

Comment viewing options

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

Thanks for the unabridged version Rik !

OddBot's picture

I wonder how hard it was for Miki to translate and condense that all down.

As I am learning, translation is hard, misunderstanding a translation is easy.

fritsl's picture

:D

Gareth's picture

Nice right up rik - and a neat list of links to steer Theo infused people into action.

At Least now the LMRians can savour what you really wanted in print....