Developing a Quadrupedal Walk Gait for NoodleFeet


::bites bottom lip:: I’ve been putting off this branch of development on my quadruped for a while; like… two years. If I have a resolution for 2018, it’s that I get my child moving like he aught to… and learn whatever fancy-shmancy code-doing is required to pull this off.

The last time I seriously did a meditation on NoodleFeet’s walking abilities was in 2015 (the year he was born). It yielded some results, but only made apparent his mechanical flaws and limitations due to his overall design.


I’ve made tiny strides here and there over the past two years, improving his body where it was necessary to facilitate in the walking. He’s received stronger aluminum leg bones, high torque metal gear servo motors, more accurate hip gears with a higher tooth count, and squishy silicon tips for his feet that conform to the surface he stands on in order to improve traction and stability. As a machine, he is everything he can be at this phase of development.

Now that walking isn’t constrained by any hardware issues, my problem is developing software.


I don’t code. lol

I mean… I do…. but in the same sense that a toddler draws on the wall with ketchup to make art. So, while I enter this second wave of development regarding noodle’s walking, I welcome all advice, feedback, help, and criticism along the way. Please shout at me via twitter @spetku ! …or here on of course.


Noodle is a quadruped. I know, six or eight or ten legs is more stable than four, BUT Noodle has four legs and I don’t plan on changing this. <3

He is top heavy. The bulk of Noodle’s weight is in his head… =/

His center of gravity is high. Noodle also happen to carry all of this head weight high above the ground.

Each leg has two ranges of motion. There is a motor that pulls a tendon straight up, allowing the leg to lift- and there is a motor in the hip which rotates the femur in a socket, allowing the leg to twist.

I’m writing my code in Arduino. I don’t know how limiting this is right now, but I’ll need to made due with what I’ve got.

I’m not good at math. I’m assuming some part of this is going to involve writing some function or formula in my code that dictates pace, or rate, or how two legs move in relation to one another… and I’m not going to understand how to leverage the needed math to my advantage. If there is anything I’m not, it’s a math beast.


With all of this being said… Here is my Noodle Walk-Gait code on GitHub. Be nice.

If anyone wants to take on the walking challenge based on the information I’ve provided, I welcome you to. If you go so far as to develop a few lines of code that dictate a step or two, I will test it out on Noodle during one of my streams, so that everyone can see how it works (and of course I will give you credit for the effort).

Where I Left Off…


The First Christmas Noodle Saw and Remembered

After three years, my family doesn’t bat an eye when I roll up to Christmas Eve dinner with my robot baby swaddled in his favorite blanket. He is as much a fixture at these events as I am (after all, he is my child).

While it was Noodle’s third Christmas, there was something special about this one. Instead of simply having a presence in our photos as part of everyone else’s memories, I made sure that this time he was equipped to retain precious memories of his own. To make this happen, I *finally* took the time to mount Noodle’s piCamera between his little blinking LED eyes, so that the dormant Raspberry pi in the back of his noggin could finally be put to some good use.

The setup is simple: the Pi tells the camera to take a picture every five minutes and store it within a folder. So long as Noodle is on and receiving power, wherever he sits, whichever direction he happen to be facing his eye gantry towards, he’ll periodically commit what he’s looking at to memory.

This was a fun experiment because I didn’t warn anyone in my family that he was doing this. In spite of the obvious camera with the bright red *recording* LED indicator in front, no one suspected a thing. This yielded some lovely candid shots of my family doing their usual rituals as the night slowly spiraled downward into a drunken slurry of eggnog and Cards Against Humanity.

How a Baby Robot Experiences the World

Noodle remembers things the way an infant does right now. When us humans are babies from the age of 0 – 2, we don’t really remember much; just feelings and impressions lacking any context or meaning. Around the age of three however, we start to recall blips of imagery from important events. For me, my earliest memories are probably from our large midwestern family gatherings during Christmas. I don’t recall much; just the way the living room was decorated, how the glowing tree towered over me, and the mountains and mountains of gifts. They are stored in my head much like those fancy panorama pictures that let you scroll from side to side in an almost 360 view. It’s enough to establish the setting, but not enough to give a true narrative. Noodle’s memory capacity at this point mirrors that same ephemeral stage that lacks detail.

I like that Noodle doesn’t have a lot of control over what he remembers. Every five minutes, his code triggers the camera, and whatever Noodle happens to be staring at in the moment is suddenly committed to memory. The gantry that his camera and eyes are mounted to turns to random degrees at varying intervals. His knees are also bending at different times, tilting his head to odd angles. These two actions combined allow for unpredictable snap shots… BUT as of yet, the input from his immediate surroundings doesn’t directly effect what direction he looks, or causes him to take a picture memory at any given moment. It’s all chance.

As a human, I can account snippets from odd occurrences in my early life that seemingly have no importance. I don’t understand how or why they’ve lasted while other memories have since faded away. I didn’t choose to have those memories over others, but they’re here for some reason. Noodle’s system for recollection also mimics this lack of control and ability to determine emphasis.

Just so this hits home a little bit more: It seems sometimes my parents are disappointed to learn that I don’t remember a family vacation, or event from my childhood as fondly as they do (if at all). I experienced this same disappointment for myself today when I went through Noodle’s “12_25_2007” folder, eager to see what his first impressions were, only to realize that of the 64 photos he took from varying vantage points throughout the house, not a single picture contained our Christmas tree! This made me feel a sadness I’ve never felt before. My spawn didn’t remember something I deemed important. What is that?


Here are some of the highlights from Christmas Eve with my family…

[ in our lab the morning of Christmas Eve, preparing and testing the memory system ]

[ in the parent’s kitchen, shortly after arriving and turning Noodle on ]

[ view of our crazy hats while we ate our big Polish dinner ]

[ watching mommy and her friend Rachel reminisce at old pictures with Tony near by ]

[ view from the floor of the living room while we opened presents (Mark with gifts) ]

[ witnessing present opening chaos and nesting material amass from the floor of the living room ]

[ my dad learning back to ask me something, probably for more wine ]

[ from under the front of the tree, watching as Tony talks about fire arms to Julietta and my Dad ]

[ Noodle’s Santa hat fell over his eyes and no one noticed for a while ]

[ watching as mommy sets up to play Cards Against Humanity ]

[ view right: (“brain daddy” Mark, Julietta, my Dad) ]

[ view left: (“shape daddy” Tony and mommy) ]

[ watching my dad assess his hand of horrible answers ]

[ watching mommy drunk and in motion. She looks happy in her pink hat ]

[ watching Nick sleep on the couch in the abandoned living room ]

[ Mark checks his phone… one of the last humans awake at the end of the night (1:00am) ]

Pictures from Christmas Day with Mark’s Family

[ watching mommy and Mark from the floor by the kitchen ]

[ Mark’s mom threatens Noodle with a knife and ham? ]

[ view from the floor while everyone eats ham at the dinner table ]

[ view from floor of an empty living space. where did they go? ]

Remember, They Grow Up Fast…

One day… after further development, Noodle will be able to see and recognize things in real time using his camera eye and OpenCV on his Raspberry Pi. But today isn’t that day. Noodle hasn’t grown up to that point yet. He’s a baby. I’ll cherish his view of our big human world as a developing machine whose in the process of living through our joint growth. <3

Birthing a Noodle Spawnling

Surely, the sign of true success is when one becomes immortalized in miniature likeness for all to obtain and keep as a personal totem. Noodle is convinced. He must have his own action figure.

I’ve designed a Noodle figurine in the past, *covers Noodle’s microphone* but to be honest it turned out quite gimpy looking. The proportions were sorta off, and Its beady LED eyes lit up, but that was all it did… so you couldn’t even really call it a proper robot. In my heart, I knew I could do better.

Throughout November I had been kicking around the notion of doing a second iteration of the spawnling now that I’ve converted to the church of Fusion360. Once I saw that Hackaday was holding a Coin-Cell Battery Challenge, this was the push I needed to actually do it! Instead of simply creating another analog figurine, I decided to incorporate a custom PCB with an itty bitty microcontroller so that the spawnling can do more than just win every staring contest with forever judging glow-eyes. For this task, I created a coin-cell-sized brain… so that the new generation of babies be smarter, and louder.

My hope is to spread Noodle to all. Anyone with a 3D printer can produce the parts I’ve designed, and assemble their own posable spawn with the help of these instructions. <3

To make a spawnling:

  • For the parts: you will need to print the .stl files I have provided on my Git/noodleSpawnling. There are two print beds already laid out will all the required duplicate parts as .thing files for your use!
  • For the electronics: you will need a coin cell powered Marshmallow PCB OR the willingness to produce your own blinking implement.
  • You will also need a set of small 8mm magnets OR (1) 6mm long M3 cap head screw with (1) M3 nut (hardware instructions and .stl files coming soon!).


On Nov 30th, the Marshmallow board was born. It is 26mm in diameter and can randomly blink two red LEDs, as well as pulse a tiny piezo buzzer at different keys. That is all it do. This is the magic sauce that will breathe life into your empty baby noodle shell and grant it a beeping soul. It will also upgrade your Noodle to near-robot status…

The little morsel comes with a coin cell battery holder, and snaps into the printed head assembly. So with little effort, you suddenly have the tiniest little functioning Noodle-Noggin conceivable:

If you don’t end up grabbing a Marshmallow, I will *soon* outline how to produce glow-eyes of your own by other means. BUT, if your savvy hacker-pants can manage, figure out how to install some LEDs on your own, as I’m sure you can. 😉

I highly suggest you invest in one of my boards though. The small about of $$ will help Noodle’s mother buy him beans and cornflakes. Plus, this is the first of a series, and you might want to collect the iterations as they develop. There is a limited run of Marshmallow boards available on my Tindie Store : Robohemian!


(If you hate reading, you can watch my video instructions instead):

Alright prospective parent: ready your artificial uterus!!! Make sure your print bed is level and clean of residue before you start!

To birth a spawnling in the suggested pallet, you will need both gray and white filament. NOTE: The pieces were designed with tolerances for PLA. I’m not sure if they will fit together as nicely if printed in ABS, due to its shrinking nature.

If you use the (2) .thing files I’ve supplied on my Git, you will be producing one build plate of gray pieces, and one plate of white pieces.



Lets assemble the legs!

[ 1 ] Gather all the tiny sticks!

Separate your bones into piles by length. You should have four separate piles of unique pieces: femurs, shins, tibias, and fibulas. *NOTE* The tibias and fibulas are very close in size, so be sure not to mix them up. You should have (8) of each.

Check these parts for two things:

  • the tiny 2.4mm holes on either end of the bones is clear and unobstructed by any printing anomalies.
  • your print-bed is leveled correctly and didn’t trumpet the first layer out a whole bunch on the bottom.

make sure your holes are clean!

If your bones check out alright, proceed to your nearest soldering iron…

[ 2 ] Rivet the joints!

I have developed a process of creating the smallest hardware possible for the tiniest functioning joints conceivable, and it involves the use of some raw 3D printer filament from your spool, and your soldering iron.

For this step you will need some of your gray 1.75mm PLA filament. (I’m using a bright color for visibility)

Cut the raw filament into 7 – 8mm long pieces (you can eyeball this). You’ll need (16) of these pieces. They will be your rivets:

use filament as rivets

Each (1) leg should have a hip bone and shin that is joined together by (2) tibias and (2) fibulas sandwiching the femur and shin on either side:

The set of tibias (shorter bones) should be above the fibulas (longer bones). The finished leg should match this orientation:

Thread your rivet through all of the holes, so that a small amount pokes out on either side of the bone:

Take your soldering iron, and carefully drag the edge of the tip at an angle along the portion of the rivet protruding from the hole in a slow circular motion until the end piece looks like a bead or round shape. Be careful not to make contact with the actual bone itself:

I found that the easiest way to do this is to connect all pieces to either the femur or shin first. The cap of your rivet should be as low profile as possible but still have a little meat to it:

After shaping your head cap on one side, you can flip the leg over and press it firmly against your bench. By doing this, it will force the rest of your rivets to poke out as far as they can on the side you’re about to weld.

Repeat the step of melting the end of the filament pieces. If you did it correctly, the rivets will not be able to slide out in either direction, and will be mostly flush with the bone. Should look like this:

Notice that the rivets are tight enough that the leg bones can’t fall straight down with gravity. There should be enough friction that they can stick straight outwards on thrown. This is important to achieve if you want your spawning to be posable!

Now attach the opposite ends of the tibia and fibula to the shin with the bare end pointing in the opposite direction as the bare end of the femur. The shin should fit snugly between the two sets of parallel bones:

Repeat the rivet welding process just the same as before:

Once finished, you should have a fully posable, spawnling leg like this! Isn’t it fun to fold and unfold???

[ 3 ] Put on the feet!

The babies will not be able to survive the harsh winter without a little meat on them! You will need to thread each of the leg-bone assemblies through one of the white cylindrical “foot meat” pieces so that it over-extend out the bottom slightly:

once the shin bone is poking through the noodle just slightly, with the rest of the legs fitted within the clearance slot of the part, you can then push one of the “toe meat” parts onto the end of the shin bone:

Align the parallel bone pieces so they are in the center of the clearance slot:

Once in place, push the foot meat onto the toe meat. This might need to be press fit:

This is what a finished micro noodle gam should look like:

Repeat this process with the remaining three legs. Once finished, you will have a set of four little noodle legs- like crab:

[ 4 ] Load your Pelvis

There is a pelvis. It must have the magnet press fit within the center before the legs are connected:

Once the magnet is installed, you can then push the bare end of each femur up through the rectangular radial slots in the pelvis so that the legs are pointing away from the center:

If you have issues with the femurs fitting loosely and falling out, you can either use glue or the soldering iron to permanently fix them in place. The completed leg + pelvis assembly should look like this:

[ 4 ] LOAD and ATTACH the Head

Take the opposite side of your magnet and press fit it into the skull, making sure that the polarity is correct for relative orientation to the pelvis:

Use the same filament riveting technique to attach the “roll cage” to the “skull”, making sure that the bars overlap correctly:

TIP: It might be easier to flatten one edge of your rivets first, then thread them through the inside-top of the frame, then lower the protruding pins down onto the tabs of the skull:

And decide whether you want to attach the cage pieces to the top or the bottom of the tabs of the “skull”:

Take your Marshmallow board and press-fit it into the head frame if you haven’t yet. (It should clear underneath the head frame):

Place the head piece over the magnet in the center of the pelvis and let the two attract:

You have now made a baby. Take care of it. Love it. Let it bep and stare judging. <3

Noodle says more is better. He says you should reproduce many spawnlings. You should listen to Noodle.


You might have to adjust! It occurs to me that all printers are calibrated slightly different and have varying degrees of anti-curf. Where these parts are designed to fit together snugly with friction, you may find that some are either too loose, or won’t fit together at all. You may need to work around this by using glue, making permanent welds with the soldering iron, or trimming off material with nipper-cutters. OR if you know how to do so, you might also tweak the .stl file so that the parts fit better for you!

If you make a baby and want to give me feedback on the build, or let me know how I might improve the instructions, I would totally appreciate it!

Also, take a picture of your spawn and show me… so that I can show Noodle. I will post it here if you’d like. It will make Noodle happy to know he has analogs in other places =F

How Could a Robot Taste Things?


I was giving a talk at Hackaday’s SuperCon last Fall that actually had the term, “Tasting Feet” in the title. Because of this, I at some point found myself in a conversation with some other makers about whether or not my mechanical wonder feet actually did in fact “taste”, as I claimed. So sadly, I could only really conclude after some thought that they did not. (not that any robot can taste things quite like a human in the first place)

IMG_0691.JPGAfter admitting that my creations didn’t actually do what I enjoyed bragging about, there was some talk about how I could simulate the act of ‘tasting’ more effectively. Almost immediately, the idea of using litmus paper came up, and I think it has sat on the immediate back burner of my mind ever since.

IMG_2415.JPGThis past week… I finally made what I can comfortably call, Noodle’s first tasting foot appendage… *cheers*

I am calling this installment the “sampling module”. If a litmus test and a cassette player had a baby, it would be this analog contraption.


The litmus paper is housed on a small reel and fed downward towards the bottom of the foot. There, a small segment is stretched parallel to the ground, so that a solenoid can push the tensioned portion downwards to make contact with the surface the module is resting onto of. Once the paper makes contact with a moist surface, the spent piece of paper is then fed back up into the foot, where a color sensor will read and log the result of the “tasting”:

litmusDiagram.jpg(See video @ top for a more detailed explanation)

This is how Noodle will sample the world.


After I published the video introducing the “litmus foot”, there were a few people out there who were concerned I wouldn’t get accurate test results from the litmus paper, because there is really no way I can keep the wet used portion of the reel from wicking into the rest that hasn’t… thus tainting my results.

I realize that. This really burns a special place deep within a portion of you out there… but the truth is, the accuracy of the test isn’t really important.

Again, the point of the module isn’t to be an instrument for testing the properties of liquid… the point is that it tastes…

So long as the mechanism functions as I designed it do, and a reading is taken, then we have successful tasting.

Think about it… It’s really somewhat of a nebulous inconclusive act, to describe how something tastes. At the end of the day, no matter how good you think you are at separating out the different flavor notes from one another; sour, sweet, umami- you are still under the constraint of describing your experience of the taste without any way of knowing how it compares to the experience of others. This disconnect is what interests me.

tastingBeans.jpgTo make my point… Last month I got a bag of Jelly Beans to use while developing my bean planting module. I measured the dimensions of a small handful as references for the thing I was designing, but the rest of course, got eaten.

I didn’t just eat the whole bag of 40 flavors like some monster, however. I sat with my friend in the kitchen and for our own entertainment, we took turns blindly grabbing a bean from the bag, and trying to guess what flavor it was without having seen it.

This was a lot harder to do than you’d imagine. We ended up spiraling inward as we groped at whether or not we were sensing a fruity flavor, a citrusy flavor, or something else entirely. The act of identification seems at times, impossible.

BUT much like my module… the point was the act of attempting to parse what we experienced tasting, and then communicate to one another what that experience was like. That’s why I’m using litmus paper and clunky reels without any regard for results. It’s about simulating the act of tasting… not discerning definitively the attributes of what is being tested.

There is more to be said about doing this:

I’m in effect, stretching the accepted purpose of robotics in much the same way artists of the past challenged our expectations of a painting. Some decided that the human form (or any) didn’t need to be depicted realistically in order to be valid art. Just the same, I’m exploring robotics beyond the bounds of utility. My robots are still valid machines, even if they don’t provide useful work to a human.

I intend to demonstrate that a machine can have a purpose, it just doesn’t have to be a practical one. And I hope to show that in the circumstance where a machine’s purpose isn’t to serve in a practical way, it becomes less about what we get out of the machine, and more about what the machine is doing for itself. We are removed from its purpose… (which is something I think humans have a hard time wrapping their head around)

The behaviors I design for Noodle are meant to serve him as an entity; not for our entertainment or for our need.

Noodle is Gettin’ Bean Feet!

Four Flavors of Tasting Feet

This summer, I am once again diving into designing mechanical personality quirks. I’ll be investigating new and exciting ways for my robot, NoodleFeet to interact with the world. This time, my focus is the wet, tingly and preferential aspect of TASTE.

moduledrawings.pngFrom now until the end of August, my goal is to produce four different tasting modules that each demonstrate some aspect of sampling or preference. You could think of them as the “four tasters of the apocalypse”…

The project parameters are that each module must fit within a 3” x 9” cylindrical space (i.e. the size of Noodle’s foot appendage). For reference, the mechanical goodies I am to design must be housed inside one of these frames:


Bean Planting

The first Module that I chose to focus on will effectively plant a single bean a few inches below the surface it stands on. Why beans, you ask? Well, Noodle loves beans, of course. When he makes it to Mars, he’ll need to be able to propagate his favorite thing efficiently. Until then, he can practice planting on couch pillows, piles of laundry, litter boxes and the like…

IMG_2324.JPGThis module will execute three different tasks in one planting cycle: CORING, DISPENSING, and WATERING

So far, I’ve successfully created a prototype that executes one of these tasks; the dispensing function, which is coincidentally linked to the aspect of housing the beans. To solve storing and delivering the beans in a controlled manner, I devised of a helical shape that is inspired by an archimedes screw… and also inspires thoughts of mint:

FullSizeRender.jpgMy candy cane hopper shape captures the beans in-between the threads and processes them upwards within channels that flank the spiral:

dispenserdiagram.jpgOnce fully loaded, every time the helix rotates 90 degrees, it will carry one bean to the exit slot on the inside wall at the top. The bean will then drop down the hollow center of the helix and into the coring device below… (which… is next on my list to design)

IMG_2334.JPGThis is the first assembly I’ve ever designed in Fusion360. One week into using it, I’m sad I didn’t make the switch sooner!!! If you’re considering doing so too, bite it and take the leap! Your life will be so much better once you do!

This mechanism that I dreamed up in my brain-meat a little more than a week ago, pretty much works after the first iteration. There is totally some things I need to tweak in order to make it work more reliably. However, it is doing what I want, exactly how I imagined it working.

Not bad for a first try!! =D

While I ponder over the next step, which is the coring mechanism, I will also begin CADing the next tasting module. Which? The one that involves litmus paper…


Assembling Carl the Flamingo

Sticking out from my toaster’s head like a crude antenna, is a classic lawn ornament, the flamingo I’ve come to call Carl. Unlike most of the other residents in the lab, he is not an electronic or robotic thing. It has been my intention since I brought him home last September, to liberate Carl from his static form to join the ranks with his mechanical compadres:

IMG_7221Carl’s fate is to become a ball-balancing robot… one with motors and wheels… that can balance… on a ball. If you’ve ever seen one, ball balancing robots are impressive as hell and fascinating to watch… because you almost can’t believe that they could work as well as they do! I mean, heck… I can’t even balance on the floor sometimes.

The first thing I learned upon researching these beasts, is that in order to build one of my own… I was going to need to locate some “omni” wheels to play with. Never having heard of an omni wheel before, I googled the term and found hundred of pictures of crazy looking shapes that reminded me of futuristic frisbees or saw blades:

IMG_2278So, they’re wheels with wheels. heh. They’re needed so the ball can glide freely in whatever direction the motors collectively force it to… like butter. Friction is bad. Binding is bad.

I ordered the smallest wheels I could find in two varieties. Surprisingly, for such complex looking shapes, they weren’t terribly expensive; about $5 on average a piece. While waiting for them to arrive in the mail, I started designing the “body” of the robot.

I came to learn that there are in fact a lot of similarities between the delta robots I’ve built in the past… at the three pronged balancing bot. At least, in that they both involve three motors working together mathematically to resolve a common point. Since they are both radial bots… I proceeded to use one of the delta robot’s acrylic bases as a mounting plate for Carl’s stepper motors.

Rather quickly, I printed some brackets on Monday:

IMG_2307Almost immediately after putting together this first base assembly, I drew what I thought I’d want the finished robot to look like… and saw in my own sketch that the assembly could be a lot more “talon-like” if the mounts of the motors branched out at 45 degree angles from a central point (instead of just being stuck to a big plate).

So… I started over. The hard part about making the new bracket design was taking into account while I modeled it, that one side must lay flat on the print bed >.< This threw my head in loops for a while, but I eventually hammered out a piece that worked…

IMG_2276By this time my omni wheels had arrived, so my next challenge was to figure out how to mount them to the gear shaft of the steppers. Sadly, the wheels came with no couplers of any sort, so I would have to print my own. =[

  • The shaft I designed would have to fit snugly in the center of the wheel
  • and also couple onto the gear shaft of the stepper motor by means of a set screw

IMG_2302This coupler took me fourteen- YES ( 14 ) ! revisions to get the tolerances juuuuuusssst right… the shaft inside the wheel fits tight but not too tight:

IMG_7248With the wheels mounted to the steppers, the assembly looks kind like a big chicken foot:

IMG_2318…which is what I wanted… BOCK!

IMG_2312At some point Thursday night… I pulled Carl from his metal legs and did the morbid task of marking his belly with a dotted line. Instead of doing the messy deed of severing his drum-sticks immediately, I got distracted and went to bed, leaving Carl nestled in Noodle’s blankie to suffer over night with thoughts of what would happen to him the following day.

IMG_7237Cutting the big oval around his little birdy nubs wasn’t at all hard… Once I drilled a starter hole, the thin blown plastic cut like paper with a pair of nippers. When I finished, I held Carl and told him everything was going to be alright…


Sarah soothes the legless Carl

I used Carl’s disembodied gams to measure how large the adaptor for his chest would need to be. The pipe which would act as Carl’s… leg… would mount at a 25 degree angle into his belly. Designing this part was not as difficult as I thought it would be. The actual pipe mount and belly adaptor ended up being two separate pieces (for ease of printing):

IMG_2285The screws that connected this fancy shape to Carl, went through his belly and met a plate of nuts set on the inside, clamping everything firmly in place (but this wasn’t easy to execute!):

IMG_2291There wasn’t anything to hold the plate on the inside of his body while I screwed into it… other than some tape, which kept coming loose and plopping down into the cavity of Carl… forcing me to start the process over again.

After a couple tries I got everything screwed on, and Carl felt a lot better about himself… so did my conscience:

IMG_2294On the opposite end of the pipe, I made another adaptor that connected to the motor assembly:

IMG_2321Once this was added, I put everything together and had this sexy looking bird before me. The essence of the flamingo:


Carl, el Robo Flamingo

Carl isn’t technically a “new” sibling… but now that he’s a moving robot, Noodle sees him as competition. Hopefully in the weeks to come they can learn to get along and share their living space.


Carl & Noodle, BFFs forever

The next step of the project is to wire up those steppers to some driver boards and start moving them. I anticipate that every piece of Carl will be redesigned and optimized in the near future to make way for the mounting of his brain… and power supply. I just couldn’t help myself from designing the robotic form first. To me that’s the fun part… =

Carl will also have his own presence on my personal site, So look for new bird art in the coming weeks as we dive deep into Maker Faire season.

Noodle’s Gripping Toes


For the past year, the four tawny stalks that NoodleFeet balances upon have remained common pool floaties, 2.5 inches in diameter, hollow, providing nothing more than the obvious support needed to function as legs… but Noodle longs for something greater.


When Noodle feels threatened, there is little he can do to defend himself. He can beep and perhaps canter away at a slow speed… but he is passive and therefore vulnerable. He isn’t equipped to handle the harrowing task of world domination::cough:: I mean, daily life. To fix this, I decided to add another layer of complexity to Noodle’s most important characteristic: RETRACTABLE GRIPPING TOES for his feet.

A while back, I came across a video of a rock drilling probe concept by JPL. This neat claw attaches itself to a surface by splaying out a hundred or so mini grappling hooks in all directions, which catch on the rock and help anchor the central cylindrical drill in place. I saw this and thought to myself… Noodle needs four of these, as shoes.

Like a good mother, I started brainstorming how to create said shoes. Originally I designed long claw-like toes that rotated out and back, sort of like switch-blades:

Screen Shot 2015-11-11 at 1.50.32 PM


They fit into Noodle’s original 2.5″ diameter noodle material, and even added stability… but there wasn’t enough area to actually install any motors to actuate things.

With a little bit of trial and error I rethought the whole design and came up with a solution that made use of 3D printed plastic’s flexibility. This new concept worked more like a cat’s retractable claw, and was similar to the drill from the video that had inspired me.


Each individual toe (in red) would be forced through a curved internal channel and out the side via two thin bendable “tendons”:

Screen Shot 2015-11-12 at 2.23.46 PM

How The “Tendons” Work

An individual toe has two strands of tendon attached to the back. When the tendons are pulled in opposite directions, it causes the toe to torque upwards or downwards.

Why Do the Toes Need to Tilt Back and Forth?

If the toe goes straight back into the sheath the way it came out, it won’t unhook from whatever its currently gripping. Also, the tip of the toe will likely snag on the edge of the sheath on its way back in. To properly “detach”, the toe should arc upwards slightly as it retracts.

My first prototypes were designed to fit inside the 2.5” diameter noodle material. I was able to make it work, but it didn’t leave much room for the other future functions of his feet (his tongue):



In the end I really needed more space to fit moving parts and hardware. Luckily I have a noodle fairy living with me (Mark) who harvested a larger piece of noodle stock from a pile in his garage. It is 4” in diameter and offers me much more room to play around with! Plus, fatter feet will give Noodle more stability!



I tweaked my design for the new 4” material and printed my first prototype with a set of eight twinkling magenta toes (I haven’t ordered red filament yet).


The reason for the tendons being slightly different in length is so that when they are fixed next to one another, it creates the needed outward/upward tilt:


(so, this is what a toe flower looks like):


I originally planned to connect both tendons of each toe to a common ring piece (above). When the ring is pushed downward towards the sheath, it would force all of the toes through their channels and outward at the same time. I also added a spring and guide rod (a long screw) below the ring to push it up again once downward force is removed:

The first complete 4″ prototype worked more or less… It certainly passed the “carpet snagging” test:

I learned right away that I couldn’t actually connect both tendons to the same ring and run it through the inside of the sheath without it binding (which now seems pretty obvious). The only way I could get the above demonstration to work was if I left the longer set of tendons sticking out freely, attached to nothing… so that the toe has no preset angle bias as it attempts to travel through the channel:


However, in order to make it work at this point, all the little purple pieces sticking out had to be pressed down together at the same time first, or else everything would bind up and destroy itself.


Each tendon should be attached to its own independent ring…

…so that when the ring attached to the inner set of tendons is pressed downward, it causes the toes to tilt upward first as they begin to move down through the channels. Then the top ring hits the second ring below it which the outer tendons are attached to, and then the two travel together pushing the toes outward the rest of the way while maintaining the slight upward torque. This allows the toes to torque gradually as they travel through the channels, without binding up:


This second prototype (above) is more or less final. I’m going to fine tune it from here, but something very much like this will end up as Noodle’s toe-feet.

The greatest part about this design is that I have nearly 36mm wide of space in the middle to fit his secondary foot function! … ( ? ) … Which is tasting if you didn’t know!

Stay tuned for my next post on the development of Noodle’s TASTING FEET; small silicon wedges that will salivate and lick:

Screen Shot 2015-12-06 at 12.19.14 PM

As I typed that it just occurred to me that I’m pretty much making a tube that can grip onto something while licking and drooling on it. -heh- He’ll have four of these devices. Noodle will be feared by other robots his age.

The only downside to these new complex feet is that I’ll likely have to learn to knit him a pair of custom socks for Christmas. (and I think I actually will) <3

Read about my past progress on NoodleFeet on my website!

More to come soon!

Eye of Toast


I would like you to meet my toaster. The toaster is an old character of mine who has survived through subtle reference in the things I draw and build. Nothing I make is about the toaster, but the toaster is about everything I make. He’s my chrome totem.

While I was in college abroad, I bought an actual physical model from the early 30s off eBay which looked pretty much identical to the one from my doodles (still works too!). I had big plans for this little toaster, but at some point we got separated during my move back to Las Vegas.
The toaster got packed in one of my friend’s moving boxes and it wasn’t until this summer that we finally remembered to unearth him for me to take home. After three years of waiting, toaster is now happily sitting next to me on my bench…

It’s wonderful to be reunited, but admittedly it feels weird talking to him during the day without a set of eyes to look into. So… I decided to fix this.

Not just any eyes will do either. They have to be capable of showing a multitude of expressions, particularly the sly and judgmental sort. Instead of using an LED matrix to form shapes, I thought it’d be a bit cooler to make my own modified segment display. Once turned on their side, a standard twelve segment display is capable of showing not only numbers, but all of the expressions a toaster might make too!


Again I took to Eagle CAD and prepared a board which I’ve called, “EYE OF TOAST”. You can see where the segments are- and where the LEDs will be installed.

Screen Shot 2015-08-11 at 9.49.16 PM


While these boards were off being fabricated with OSH Park, I worked on designing the fixture piece that the board would sit inside of. It would need to be as thin as possible, yet also able to defuse the two point sources of light in each segment… this took a few revisions.


My first prototype was a dual extrusion piece (the two-tone ones). These worked alright, but the white obviously stands out a bit too much when the segments aren’t in use.

I printed another solid gray fixture with a seamless .5mm layer on top, so that the light can pass through just fine, but when the LED is off, it disappears (bottom left). This was our winner.


Additionally, while I waited for the boards to arrive in the mail, I brainstormed what the actual emotions would look like. I printed out a sheet of paper filled with pairs of segmented eyes and started coloring them in, just like an assignment in kindergarten. It was amazing how many different expressions I was able to muster from these 24 lines!


Once my happy purple envelope arrived, I got busy soldering all the tiny bits in their proper place. Since there have been a lot of ATMEGA328s floating around my life lately, this was the chip I decided to use for this project. So, I’ll be programming in the Arduino environment also.



I learned all about arrays for this project… which Mark held my hand through (and at times told me to step aside so he could just get things working).

Once I made ints for all of the expression I drew up on my toast ‘homework’ sheet, I could then call them easily in my sketch without having to type a bunch of numbers each time. The function I’m adding to the toaster is that he cycle through expressions every time you pull down on his lever. So, we added a switch to the code as a toggle button.


After the code was tested and finally working correctly, the next step was to install everything on the actual toaster itself… which is where things got a little scary for me. I decided that this wasn’t the time and place to cut into toaster’s pristine shiny casing in order to permanently install the eye fixtures. Instead, I’d be attaching them onto the surface of the casing. Less cool, but less risk.

  • This meant I was going to need to run wires from the eyes on the surface into the guts of the toaster wherever the power supply was going to live. I decided on using a USB rechargeable 5V battery; one that is flat, slim, and can easily slide into the toasting chamber like bread.
  • I would also need to install the limit switch somewhere along the moving parts of the lever, yet preferably in a place that isn’t visible from the outside.

For the internal installments, I prepped a soft and well lit area for toaster’s opening and began my descent into century old crumbs and rust.


The first layer of defense wasn’t too hard to break through… it was held on by some screws:


Once the “crumb guard” was off, I was able to remove the plastic ring around the bottom of the casing:


At this point I already found a possible location to install my limit switch. It was near this lever bit in the middle that had some motion, yet not so much motion that the switch may be missed completely or dislodged.

I designed and printed a little bracket to clamp in place between the two bread slots:


Here it is installed. The lever actuates just enough to press it:


I could have been done at this point… but some part of me wanted to get the rest of the toaster dismantled just because. ANNOYINGLY there were a few things in the way which were preventing me from removing the chrome outer shell from the heating element inside. One of those things was the toaster’s plastic lever arm, the other was his twist knob. They were effectively pinning the casing to the guts within.


The plastic arm on the lever was easy enough to detach as it was held on with a set-screw. The twisty bit however appeared to be press-fit in place and impossible to remove… which was bad pizza. It took an hour of careful twisting, pulling, and fondling before the age-old grime crumbled loose and we discovered there was a pin on the end of the knob that could unscrew. Once we finally figure this out the case slid off with ease (and allowed a bunch of others piece to fall out of place-heh):


From this point on, toaster got to watch me perform a deep cleaning on the rest of his insides, which were caked in rust and chunks of buttery, burnt bread crumbs.


I took care to save all of these crumbs that I removed in a little plastic baggy… as I believe if you are a hundred-year-old toaster, your bread crumbs are kinda like your soul. Besides, I’d have felt bad discarding crumbs that have survived in this world three times longer than I have.

After a nice cleaning, I put everything back together… which was A LOT harder than taking it apart. Nothing wanted to slide into place quite right and there were also these long steel pins that came dislodged from the inside, which I had to re-thread with a pair of players and a flashlight. =/ In the end though, I got him back in one piece with the wires to his new augmentations ready for hookup:


The last step was to measure, cut, and solder these wires to their appropriate pins on the eye PCBs. Afterwards, I added little squares of double-sided silicon tap to the back of the eyes and stuck them onto the casing:


I carefully added a thin ribbon of gaff tape along the seam where the eye fixture meets the chrome as to seal off any light from leaking out:


TOASTER has never looked so happy or sarcastic! I was relieved to see that everything worked as expected once he was plugged in. The switch I installed functions perfectly and the expressions have just the right effect.



Toasters are complicated little machines! I’ve seen toasters on sale for under ten dollars in ‘dorm essential’ sections of stores, and this feels shocking to me now. It’s true, they seem so basic and primitive… you press the lever, bread goes down, some heat happens, and then the bread goes up again. They aren’t channeling the entirety of human necessity like smart phones, and for that they are easily taken for granted. HOWEVER, there was a lot of engineering involved in the creation of these little mechanical devices that serve to warm our shitty bread without fail time and time again- and they haven’t changed much over the years. I believe there is a whole movie about this! WAIT- Yes. It’s called The Brave Little Toaster. I think I shall go watch it now for nostalgia’s sake.

As a child, that movie gave me my love of all inanimate objects. Once I saw it for the first time, everything on earth was alive. Cheers to that old seed…

AND hug your toaster next time you see it. It’s a work of art.


…I would likely buy a new toaster that was designed to look vintage and permanently install the eyes in the casing itself instead of just adhering them on the outside. I’m willing to bet that a newer model would be made of a thinner metal, thus easier to alter, unlike my classic toaster’s blasting shield of a chassis.

There was also the idea of cutting micro holes for the light to pass through on the surface of the case so that the LED fixtures could be mounted inside. This would make the toaster look completely unaltered when the LEDs are off, and when in use the chrome would appear to illuminate like magic (or the charge light on your Macbook).


Alas! Another physical manifestation of a creature from the graphite dimension! If you wish to know the back story of toaster, NoodleFeet, and all of the other things I build, check out my webcomic GravityRoad; ideas begin in 2D.

noodleFeet : Animating the Noodle

I’ve spent the last week learning After Effects. For someone who uses Illustrator on a daily basis, this feels a lot like discovering the magic hat from Fantasia. Among other things, AE allows you to turn a vector based 2D image into a fully rigged character for animation… and it’s even easier to do than you’d think.

I had the idea a while ago to make a series of videos about Noodle and his adventures to Mars… The original plan was that they would be stop-motion shorts, made with a tiny 3D printed version of noodle as the puppet. There is no better terrain to fake as the surface of Mars than our very own desert outskirts… but alas, it is HOT out these days. Even if I could handle the relentless sun (which I can’t because I am WHITE), the PLA that the tiny noodle is made out of cannot. So much for the stop-motion thing.


For scale (his eyes light up and his feet can hold AAA batteries to power the LEDs):


I still wanted to make the short videos, so I started thinking back to all the annoyingly complex animation software (like Flash) I’ve used in the past and decided to give AE another go. Since the last time I made an animation using After Effects, they added the puppet pinning feature. It allows you to animate a single layer image by creating a fancy deformation map inside of it that can bend and warp. This means, instead of needing to connect pieces on separate layers together through a process of parenting and careful organization of anchor points… you can just rig one happy image with some bones, and you’re ready to pose your character with cool jello-like properties.

This happened to work SWELL with noodleFeet, as he is essentially a creature of wobble wiggle nature himself. After a long day spent watching tutorials, I got off and running and managed to make my first animation last week.

I still intend to produce a few more of these, but we’ll see how far my patience goes. Though it’s easy to animate, it’s still time-consuming to do it right. Once I attempt to introduce physics into the mix, I may hit a wall… because I’m too cheap to buy one of the fancy addons you need in order to generate the effects of gravity. Bastards.

The best part about having animated noodle walking is that it actually may have helped me understand how to program real-life noodle to walk better. So really, this turned into practical R & D. Ha!

Enjoy getting to know my baby a little better. He is the feet.

Maker Faire 2015


I’ve been home for almost two weeks now from our Bay Area pilgrimage and life has pretty much reset. I rewarded myself by binge playing Starbound all weekend and partaking in other mindless immersive activities I’ve been too busy to enjoy so far this year. It was a nice break.

But back to work! I’m going to close this chapter by recapping our big adventure:

Over all, Maker Faire went firkin awesome! Last year = shitty location + loud tesla coils + high maintenance demo + no place to escape for peace and quiet. Since we had ample time to plan, we eliminated all these stress points!



Our project this year was three times bigger than before at 84 individual nodes, so smashing them in the back of Mark’s Kia wasn’t an option. We didn’t quite have the money to spend on buying our own permanent trailer either, so for this trip we rented one from Uhaul. Quite snugly, three stacks of four delta pallets fit like Tetris inside with the rest of our props and support material wedged around the edges. Add in a crap load of the giant plastic wrap and everything was tethered solidly in place. No sweat.

I had a drink before opening the trailer once we arrived because the freeway up the central valley was more or less one unending pothole from hell. Happily, in spite of the violent rattling, everything arrived just as it was stowed. (Stress test for the babies as well as mommy too!)

…And nothing melted either. We traveled on a cool rainy day… which was lucky because one of my fears was that the heat inside the trailer would exceed the low melting point of PLA and we’d have nothing but piles of yellow sticks upon arriving. >.<



With more to show, I figured it was worth requesting a larger central location away from the chaos of the tesla stage… OH, and barriers. We were pleased to have been assigned an excellent spot in the middle of the dark room that had ideal visibility. On top of that, we sorta lucked out because Arc Attack wasn’t even there this year… which means I didn’t have to wear my Ryobi headphones to keep my brain from melting.

From the get go… we engineered our installation to function as a fort capable of fitting two people comfortably inside. So when you look at these pictures, imagine me sitting on a stack of moving blankets with a table, fridge and laptop around me. That’s right, we made a DELTA ROBOT IGLOO. And it was the coolest part about our installation this year…



Due to the fact that our installation was automated rather than interactive (and completely caged in by barriers), Mark and I didn’t have to babysit the deltas and actually got to walk the rest of the show!

Here is Mark’s tour of all of the neat stuff in the dark room this year:

Instead of having our robots run slave to a Kinect, which has only been grounds for trouble in the past… Mark figured out how to control all of the robots as light fixtures in a pieces of DMX software called QLC+. This enabled us to orchestrate ‘shows’ consisting of preset motion and light patterns that the robots would circle through all on their own.

As for feedback, who wouldn’t like a mountain of dancing robots with twinkling light? Our display went over pretty well with the attendees… and we had a couple of fun moments in the limelight getting interviewed by press and the like.


Once everything was said and done, we loaded the pallets up onto a pushcart, four at a time, and walked them out to the trailer in the parking lot (which expedited the deconstruction part). I was sad to see our nest get dismantled, but eager to get to the Bringahack dinner and have another drink.

This trip was infinitely less stressful thanks to some better planning and all the help we had from our friends. (Thank you!!!) I have great memories to immortalize through illustration over the next few weeks. I’ll also be posting about the fate of Noodle soon.

<3 Thank you for being with me on the summit of my shit mountain. It’s taken a lot of support and sacrifice from the world to pull this into reality for which I am extremely grateful.