Such and Such vs. Uberclocker Advance (credit: Charles Guan)

This past weekend I had the great pleasure of competing at NERC‘s Motorama Robot Conflict with my fighting robot Such and Such – built 100% at Hive76.

Though it might have looked a little boring, that was the most exciting match of the competition for me. After 2 years of on-and-off work, Such and Such, the most ambitious robot I’ll ever build, worked like a charm.

Some rough CAD layout of the bot.

And still it only barely managed to eek out a win. To put it simply, Such and Such was not designed to beat other robots. At least, not on its own; it’s actually meant to fight alongside a partner hammer-bot, but weight limits prevented our tag team strategy. Rather, the point of Such and Such was to come up with the most overtly weird fighting robot possible, and then design it and build it.

The two main features of the robot that make it kind of unique are the walking and clamping mechanisms. A great deal of time was spent in the design phase figuring out how best, and in fact how easiest to implement these ideas.

Walking

There were a few reasons I chose to make Such and Such a walking robot. First, there aren’t enough walking robots and this was a way to make the whole thing wackier. Secondly, I wanted  to increase the bot’s traction (I was previously using omniwheels to accommodate the sideways motion of the clamping action – unfortunately, omniwheels tend to have poor traction, so omni-legs seemed like an acceptably ridiculous alternative). Additionally, walking robots are allowed a 50% weight bonus, and I hoped to have a little left over to afford all this additional complicated mechanical nonsense.

The walking mechanism is basically a set of three crankshafts with a series of cam profiles that ride in slots on three legs. The legs have a phase difference of 120 degrees, so each rotation of the shaft provides about three “shuffles”.

Prototype leg; UHMW cams and spacers sandwich onto a 1/2″ Al keyed shaft.

Many walkers of this type rely on eccentric circles for cam profiles, resulting in sinusoidal x-and-y movement of the legs. That is to say, the up-and-down (y) and back-and-forth (x) position of the legs is never the same instant-to-instant. This means that the robot’s frame will constantly be oscillating up and down above the floor as it walks forward. Since both halves of Such and Such are so far apart, I thought this might result in some weird binding forces on the clamping mechanism.

I wondered if there was a way for the robot to stay at a constant height off the ground as it walked. It would require the y-motion of the legs to switch and hold between “up” and “down” positions. That would prevent any binding of the clamping mechanism, plus it sounded like a pretty cool challenge. Conversely, the sinusoidal motion from a regular eccentric circle would have a leg continuously in motion between “up” and “down”, instantly reversing its travel once it reached those extreme points. After a bit of research and some time sketching on graph paper, I arrived at the following cam profile.

Y-Cam profile.

As it rotates in the leg slot, it holds the leg in the “down” position (in contact with the ground) for 120 degrees, then transitions and pushes the leg up for 60 degrees in to the “up” position. It stays there for another 120 degrees. Then the next 60 degrees of the leg return it back to the “down” position. With a series of 3 legs and cams 120 degrees out of phase with eachother, there would always be a leg being held in the “down” position, and the chassis would – in theory – stay at a constant height above the floor.

Walking animation.

All of this up-and-down motion happens in conjunction with some back-and-forth motion to create a walking cam. Looking at the picture of the cams in the slots, it’s visible that the outer two cams are the binary y-motion (up and down) cams, while the center is an eccentric disk providing sinusoidal x-motion (back and forth). So as you can imagine, there’s quite a lot of sliding motion going on inside of these things.

First test batch of waterjetted parts.

A big part of good design, or good anything, is to know your limits. Knowing that I could not make these cams and legs with the necessary precision, nor in the quantity required, I had the nice folks at Big Blue Saw low-taper waterjet them out of 1/2″ UHMW for me. Once they arrived I buffed all the sliding surfaces to get them to a smooth, low friction finish. After a quick test to verify the cam worked as expected, I ordered the rest of the batch.

Clamping

The last version of Such and Such had a really cool pantograph-type clamping mechanism much like a scissor lift on its side. It was fun, but a pain to deal with the linear motion at either end of the linkages. Sliding joints are a little harder to pull off than a bunch of pinned joints, so I thought I’d like to avoid it altogether. I also wanted this version to have fewer links for the same amount of horizontal travel as it’s a slightly more efficient use of weight. I settled on the following mechanism:

Diagram of clamping mechanism with rear link driven by linear actuator.

Looked pretty cool to me. Only 4 links and a great deal of horizontal travel. But little did I realize at the time that there was a fatal flaw in the design… but more on that later. I kept the the linear actuator external to the bot so that each half can be as small as possible, only about as wide as the walking bits. Conveniently, there was a nice little space for the actuator betwen the links in both open and closed positions. I did a little bit of math to find an optimal-ish offset angle (α) for the actuator to drive the rear link at so that the clamping force is mostly constant throughout the mechanism’s whole travel.

Mechanism parameters, after a few iterations.

Clamping force as a fraction of actuator force over the course of the mechanism’s travel.

Construction

Construction commenced on December 1st 2012, just as I finished up my 8mm Projector Tremolo project. I made simple drawings of all the frame members and got to cutting all the lengths of 3/4″ angle and 2″x1″ U-channel that comprise the frame halves.

Angle brackets after the first day’s work.

Next, all the frame members were drilled out to accomodate brackets and bearings.

Frame members cut and drilled, ready for assembly.

After testing on a prototype leg and cam set, I ordered a full batch. One of the most time consuming parts of the project was cutting, drilling, and mounting the 48 little roller brackets onto the legs.

Leg with brackets for rollers.

Next came the delicate task of stacking the three sets of UHMW cams, legs and spacers on three shafts, and then repeating the process for the other half of the bot. I decided not to lubricate the cams since most references indicated that polished UHMW is sufficiently self-lubricating, and grease may only trap dirt and debris. I drilled some well-oversized holes to allow some play in mounting the self-aligning bearings for the camshafts – since the phase of the cams was of utmost importance to the mechanism, I let the timing belts locate the bearings’ exact position.

Frame and walking bits assembled.

After another order of waterjetted parts (this time aluminum), I assembled the linear actuator and attached linkage pivot points to the frames.

Linear actuator.

Two halves assembled and linear actuator installed.

With the frame halves and walking mechanism mostly taken care of and smoothed out, I switched my focus to the clamping mechanism.

Links cut and drilled.

However, as soon as I laid out the links in front of me, I realized I’d made a grevious mistake. What I failed to realize in the design phase is that the linkage has way too many degrees of freedom. What that means is, I could move that one link back and forth all I want with the linear actuator, but the rest of the mechanism is gonna flop around however it likes and not clamp at all! As a mechanical engineer, this is quite the oversight and really really embarassing. In my mind, I just had assumed for so long that the two halves of the bot stayed parallel to eachother as the mechanism opened and closed. In reality, there was nothing constraining them to this kind of motion. 3 weeks out from the competition I had to come up with a quick fix in the form of some secondary links and a sliding pivot in the center of the whole thing. If these parts look like an afterthought, they were!

Linkage in place, with secondary links.

After such an embarassing mistake I found it helpful to walk away from the build for a little and let my mental faculties simmer down a bit. Thankfully all the major hurdles in the build had been crossed by this point and before I even realized it I was ready to install drive motors and wire it up.

Actuator installed, wires run and ready for electronics.

Done and ready for Motorama!

I even had time to add a neat hinged battery compartment to make charging way easier. Before I even got to the competition, I had considered the build a success: it walked and it clamped. I came up with just about the wackiest idea I could, I built it, and it worked. So now it’s back to cheap, simple robots. Maybe.

Here’s another video – this time with Charles Guan‘s incredible Uberclocker Advance, during which Such and Such spends alot of time off the ground.

(Check out Mike Jeffries’ YouTube channel for basically all the fights from the competition!)

Huge thanks to NERC for putting on great events, Alan Young for his generous help with blown up motors, Josh Frisch for being a patient teammate, my Dad, Michael Jeffries for the video, Hive76, and the entire robotics community for being such a righteous bunch of folks.

And I just got done with a talk at ScienceOnTap Philly! It was a truly excellent night! Special thanks to the Organizers and also the Hivers who came out or emailed in their support! You peeps are the best.

Here are some pics via the Twittersphere. Thanks to the photographers for posting!

Texas Instruments MSP430 LaunchPad.

To that end, I did two things. First, I bought ten MSP430 LaunchPads. These things are really inexpensive, so they make great stocking-stuffers. If you don’t know what the MSP430 is already (really, we talk about it constantly, where have you been), it’s a 16-bit microcontroller with really low-power consumption needs. They run on 1.8 to 3.6v power supplies at up to 16MHz, making them quite a powerful little beast for only $4.30, which includes the chip programmer. If you were to buy the chips alone, they cost about $0.50 each, with a few different serial communication protocols built in, and requiring only a small selection of external parts (2 resistors and 2 caps if you want to do it right, 1 resistor if you’re living dangerously, and face it, at $0.50, you can afford to live dangerously). It’s something of a long-term project plan of mine to buy 100 of these and try to build a small, physical neural network computer.

A lot of people have shied away from the MSP430 because the Code Composer Studio software–based on the professional-grade Eclipse development environment–is very difficult to use in comparison to the Processing-based software typically used to program Arduinos. But luckily, someone has taken the Arduino cue and created Energia, a Processing-based editor for use with TI’s LaunchPad line of MCUs! If you’re experienced with Arduino, using Energia is a snap, and if you’re not experienced at all, it’s really not that big of a learning curve.

I prefer to call it Serious Putty

Second, I bought supplies to make “magnetic Silly Putty”. About a week ago, I saw this Instructable about kneading some Iron-Oxide powder into a little Silly Putty and jumped on Amazon right away to get a 6-pack of Silly Putty eggs and a 5 lbs bag of black Iron-Oxide pigment. Honstely, 5-lbs is way too much, but there are plenty of other things you can do with it, like make ferrofluids or your own paint, so it’s handy to have around. You will need a fairly strong neodymium magnet, but again, these things are fun enough to have around anyway, so have at it!

Making the putty is really easy. I pooled all 6 eggs of putty together in a non-stick pan. On very low heat, I warmed up the putty until it was just too hot to handle with my bare hands. If it starts to become the consistency of chewing gum and sticks to the pan, don’t worry, it will unstick when it cools down. Don’t heat it further than that though, it will start to smoke and burn. Wearing rubber gloves to give me just enough insulation from the heat and to keep my hands from getting stained black, it’s just a matter of working a large, heaping tablespoon-full of the black powder into the putty. You will need to work the putty like taffy, stretching it and folding it to blend the powder evenly into it. Once the powder is sufficiently kneaded in to the putty, it will not stain anything, so keep testing it on the back of your rubber glove to see if it leaves any marks. I then cut the putty into 6 equal chunks and shoved them back in their eggs. It took 10 minutes total. I thought about taking some photos of the process to show it off, but really, it could not be simpler.

I’m hoping these gifts will be completely unexpected and will inspire people to try something they never would have considered on their own. The MSP430s are just a really easy, cheap, fun way to get into programming, and the magnetic Silly Putty is a great example of something you can’t buy as a product that is also extremely easy to make.

About 30 minutes into making our own greensand molds, we realized that this was going to be a difficult process, and immediately destroyed several hours of work trying to get a good crisp mold for our first pour.

A simple gear was too difficult

Several hours into our class, we managed to finally get a good solid mold of a 3D printed TARDIS. We hopped in line and got a pretty good looking cast. Andrew also attempted the TARDIS with some success. He also managed to get some good casts of a wooden puzzle, including one that blew out. However, due to our earlier troubles, we decided to hedge our bets and get one more good pour out of the class before we would start wrapping up. While waiting to pour ours, I was being shown how to work the furnace by Gus, and ended up melting down plenty of scrap and helping others make their pours which was a lot of fun to be working with. The furnace was operating at about 1300 Celsius, and moving around molten metal at that temperature can be quite a thrill. We plan on working with Gus and Darla at Philadelphia Sculpture Gym on some other types of casting techniques, especially as they apply to our 3D printing. We look forward to working with them in the future, and hope you all consider taking their next Greensand class in January.

Exile: The Family You Choose: The Kickstarter Campaign

ZOMG GUYZ!

One of my first projects here at Hive76 was the Burning Zombie Dummy. A friend of mine had called me, asking me if I knew how to set people on fire safely, and that led into a very stern discussion about what he was trying to do and that I would take over so that noone would get hurt. So I became the Special Effects Design Engineer for Exile: The Family You Choose, and it was one of the best times of my life. I got to do some pretty awesome things (including making an impromptu harness for doing a shotgun-to-the-chest effect), met a lot of really great people, and learned a lot about a hobby that would ignite my passion in a way I hadn’t felt in a long time. Seeing our end result, this thing that we made together, from start to finish, without any adults (of course, we’re all adults, but you never really feel like it) helped to further cement my belief that anyone is capable of doing anything. The hacker spirit is strong in the indie film world.

We went to the GenCon Indy 2012 Film Festival in Indianapolis, IN, and I got to see a whole new world. Indie film crews from all over the world (yeah, even one from Australia) showed up for nothing more than to meet each other and share tips. Stories of hard work, ingenious hacks on tight budgets, and caffeine addiction were easy to find in this extremely welcoming of communities. I also got to hang out more with one of our writers, to know him better, and peek into the vision of where the series is going. It’s pretty exciting and can only get better from where we already are. A lot of dreams were expressed of building a community around independent film and theater, producing content for free on the Web, for the people, to hopefully one day break the oligopoly of big-time, Hollywood studios. Did I mention that the FOSS spirit is also very strong in the indie film community?

Today, I am the Associate Producer for the series. This is kind of a catch-all title that makes sure everyone is on task and getting what they need. Over a decade of working in software development, much of it as the technical lead on a team of developers, has prepared me for this quite well. It’s much the same sort of work: keeping notes, communicating well, and following up with people. I’m also doing as much as I can to stump for the series; we all do. Independent film is a lot like a hackerspace in that way; everyone is in it together, doing all of the work completely for free, just because we love it, pushing on through setbacks just because you think you might, just maybe, possibly can pull it off.

Your donation will help us buy the appropriate books for making films, instead of pretending that this thing was at all useful.

Everyone is very dedicated to the project, but at this point, we’ve reached the limits of what we’re capable of doing out of our own pockets. So this year, we’ve setup a Kickstarter fund, to appeal to the charity of the Internet, that maybe they are eager to support people trying to do A Thing, trying to learn, trying to grow, and maybe even invest in a mindshare of local artists. With this Kickstarter, we’re looking to improve our equipment and effects budget. For example, last time, we were trying to light a scene in the middle of a horse pasture with extension cords from a house 300 yards away. You can imagine the terror in the director’s eyes every time one of the extras kept tripping over those cords.

Fake blood, duct taped to her HAND!

Even as little as a dollar helps. Man, I don’t even KNOW what costs a dollar anymore. Half of a hotdog? The first 5 minutes of the latest Transformers movie? Dryer lint (maybe)? Just think, you can help make the dreams of a bunch of nerds and geeks come true, while also supporting the arts and the Internet media revolution, all for the price of 1 suit from a pack of Bicycle playing cards.

The original ALF 120lb combat robot, 2005.

Many folks surely remember the days of fighting robots on TV: Battlebots, Robot Wars, Robotica, etc. And while its televised days are behind it, the sport is kept alive by groups of builders and competitors across the country. The Northeast Robotics Club (NERC) is just such a group, and one that I have been a member of since I first saw robots destroy and get destroyed on TV.

In the years since, I’ve traveled up and down the East Coast competing with robots of my own. But this past weekend, our own city of Philadelphia hosted NERC’s annual Franklin Cup, held in conjunction with the Franklin Institute. For this event I decided to continue the lineage of a long-standing family of NERC bots: ALF!

Now, this is the first robot I’ve built since I’ve moved to Philadelphia. My center-city apartment is about the size of a large phone booth and lacking any machine tools, so it’s obviously not a good workspace. The Hive, on the other hand, with its storage space, large work areas, and 24hr availability of tools and resources, was the perfect place to build. It may sound like a shameless plug, but the story of ALF would be incomplete without it.

From the start, this robot was to be a testbed for a some new ideas. Firstly, I decided to make the robot an exercise in ease of construction (Design for Manufacture or DFM is what they call it in the real world). Using off-the-shelf mechanical components from McMaster-Carr and SDP/SI helped with this immensely. Standardized hardware meant that the whole machine could be stripped bare using a hex key set and a socket wrench.

Most of 30lb ALF’s innards.

For the weapon, I chose to build a quick firing spring-loaded hammer – something a little on the exotic side as far as combat robots go. I had never built anything like it, and I’ve always found something innately appealing in the cams, rollers, and linkages involved in such mechanisms. The rotary cocking/release interface was very fun to build and watch in action, providing about 180 hammer blows per minute.

One-piece Lexan top armor, complete with neccesarily obnoxious paint job.

Additionally, I wanted the layout of the robot to be very open and easy to repair. While this was helpful at the competition, the layout made for a lot of unused space inside the bot. Using a one-piece shell of formed Lexan for armor meant that every component was immediately accessible with the removal of four bolts.

In not much time at all, ALF was ready to fight! I had a simple, straightforward robot driving around with a working hammer mechanism – all of this a few days before the event and weighing in at a cool 29lb. With that I considered the build a success! I was off to the Franklin Institute to see how it would fare in the arena…

This just goes to show you that after years of building and competing, anything can happen once the match starts. Unfortunately, the motor powering the hammer mechanism stalled within the first few seconds of the fight, causing the last stage of the gearbox to fail catastrophically. That is to say, there were stripped gear teeth everywhere when I opened it up. So that was it for the hammer, unfortunately.

ALF patched up alright!

As for the top armor, I was surprised to see Lexan crack so easily as I’ve always known it to have excellent impact resistance. I tend to think that the lack of serious structural members supporting the shell allowed it to flex just enough to the point of fracture. ALF’s underside, which took a good deal of hits as well, was the same material  but bolted directly to the frame’s steel base.

But even though basically every major component inside of ALF (drive motors, batteries, controllers) took multiple direct hammer hits, somehow, the drive system worked as well as ever. For the remainder of the competition, ALF was weaponless, but there was nothing to do but fight on.

Without its hammer, ALF was a sitting duck in all its other matches (the videos of which I will leave to the reader to find on their own…). One highlight was ALF’s fight with the pneumatic flipper and defending champion, Upheaval:

ALF at the end of the competition.

At the end of the day, ALF was beaten, battered, and relied on much more duct tape than the original design called for. Ultimately though, I accomplished nearly everything I wanted to with this robot - apart from beating other robots. I can’t say I didn’t gain a whole bunch of knowledge though. And as with every project, I’m just cutting down the number of mistakes I have to make in the future.

Watch this space as I begin work on rebuilding my flagship robot, Such and Such: a walking, clamping, elaborate beast of a machine.

Big thanks to NERC, the Franklin Institute, and Hive 76.

Interested in fighting robots?
NERC - Official website, look out for competitions.
Robot Combat - Everything you need to know to get building.

Double-acting PVC hydraulic cylinder and control valve

The pressure in your typical garden hose is nominally around 40 psi or so, so my first hydraulic cylinder should be able to develop about 125 pounds of force if it had really good seals. This is a proof of concept so I didn’t bother with o-rings or anything, so it leaks like crazy and thus is unable to develop quite those kinds of pressures, although it is quite strong. Moving from a 2″ to a 3″ hydraulic cylinder would bring this up to about 282 pounds of force, not too shabby for garden hose power!

The hydraulic cylinder is made of standard PVC pipe (2″ for the cylinder and 1.5″ for the ram), although I had to use my lathe to turn down a 1.5″ pipe cap to fit inside the outer cylinder. The control valve is made of 1/2″ CPVC fittings and tubing, with the exception of the spool which is a length of 1/2″ solid PVC rod. I had to turn down the spool on my lathe to the appropriate profile and also had to drill out the valve to fit it. The fit is fairly poor but it shows that the concept definitely works. Eventually I am hoping to be able to have all the custom parts injection molded to get the unit cost down cheap enough that it would make a good toy for DIY doodlers and budding engineers everywhere.

Why be so public about something so secret? Because this tile uses a layout technique that lets you build charlie-plexed LED arrays quickly and cheaply — and that’s something worth sharing.

Charlie Tile Circuit

You need to flip the tile over in order to see what’s special about it. Here’s a quick list of features that make the assembly what it is:

• The back of the tile has six “column” conductors and six “row” conductors.
• These column and row conductors are connected along the diagonal of the row/column array.  At all other points in the matrix, the row and column conductors are isolated via a layer of masking tape.
• LEDs above the diagonal have their cathode connected to the conductive row immediately above the LED.  LEDs on or below the diagonal have their cathode connected to the conductive row immediately below the LED.
• All LEDs have their anode connected to the column that is to their immediate left.

The resulting circuit allows you to individually address any of the thirty LEDs in this 5×6 matrix using only six lines from a micro-controller.

It turns out that this assembly technique is pretty quick.  It’s also physically robust.  Best of all, it’s cost effective — copper tape is relatively inexpensive and masking tape is downright cheap.  To get some perspective on how this compares to other assembly techniques cost-wise, go price a custom printed  4.5″ x 5.25″ circuit board.

This technique can be used to assemble arbitrarily large tiles.  In fact, the larger the tile, the more effective the technique is with regard to saving time and money.  And, as a fortunate coincidence, most peg-boards happen have 5mm holes — so that means you buy the peg-board at the hardware store and jam LEDs in there perfectly w/o any drilling whatsoever — so just buy the panel, tape it up, jam in the LEDs, solder the leads and enjoy the show …

One of our core members, Jordan Miller, has just published a scientific paper using RepRap 3D printing technology to engineer living tissues for regenerative medicine. I’ll give you a rundown of the science and a step-by-step guide of how Jordan got to this great spot in his career. Jordan is quick to point out that this is work that would not have been possible 5 years ago, or without the help of RepRap, Hive76, and this wonderful city of Philadelphia.

There are other labs around the world that are attempting what Jordan and the rest of the team at UPenn and MIT have been working towards. The end goal of regenerative medicine research is engineered tissues and replacement organs for treatment of human disease. As Science news says,

Imagine a world where if your heart or kidneys failed, you wouldn’t have to endure an agonizing, possibly futile wait for a donor whose organ your body might reject. Instead, a doctor would simply take cells from your own body and use them to “grow” you a new organ.

Other lines of research are attempting to 3D print directly with living cells and gel. These so-called “bioprinting” approaches involve loading cells and gel in syringes to be used as feedstock to create a structure from scratch. The problem is that healthy liver cells, for example, usually die of starvation (lack of nutrients) and suffocation (lack of oxygen) while enduring the slow 3D printing process.

Jordan’s 3D printed vasculature approach was inspired by whole organ vascular casts like this one.

Enter Jordan and his innovation: since vasculature provides the lifeblood to resident cells, why not focus on the vasculature first?

Jordan and the rest of the research team at UPenn and MIT have developed a new way to create vasculature for living tissues. This 4 step process involves: 1) 3D printing a network of sugar filaments, 2) surrounding it with living cells in a gel, 3) dissolving away the sugar to leave behind a vascular network for 4) the delivery of nutrients and oxygen. He accomplished this with a custom built 3D printer, extruder and control software.

Here’s a step-by-step of Jordan’s many year process:

1. Get a crazy idea to link sugar and vasculature when comparing the interior of a 3D print to a capillary network.
2. Get a PhD in bioengineering
3. Move to Philadelphia
4. Join a hackerspace
5. Get introduced to 3D printing, MakerBot and RepRap
6. Assemble your first MakerBot
7. Invent a heated build platform to dry your sugar while printing.
8. Add a heater to the Frostruder so you can print molten sugar.
9. Assemble a customized RepRap Mendel that fits your new extruder.
10. Get help from your hackerspace to properly control your pneumatic extrusion.
11. Work for months perfecting recipes and methods for printing vasculature.
12. Write it all up in a research paper and submit!

You can read the Penn press release about this awesome science, an overview from Science News, or the full paper. A more detailed post about the hardware used in this project will follow and soon you’ll be able to make your own sugar extruder. (It prints chocolate too!)

And – a lot of my friends here have great ideas for DIY kits, but they don’t want to take care of sourcing, shipping, collecting money, etc. Who can blame them? There’d be many more interesting kits out there if someone solved the sourcing problem.

That’s why I launched Kitify a few weeks ago. Kitify makes it easy to document and list a bill of materials for a DIY project (a little like Instructables, but you have control over the presentation), and with one click you can also sell your project as a kit that we put together for you. You tell us what’s needed to build the kit, we sell kits on your behalf, and you get paid.

Kitify was fun to build, I had to learn quite a lot to get it off the ground. Check it out! And if you’re interested in selling a kit, let us know through our contact form and we can either help you get it set up on Kitify, or give you other advice on marketing, logistics, design, and lots of other areas.