Friday, September 5, 2014

Bug Love Robot: MegatRON for Dragon Con

I sense a change in the winds! Seriously, its getting cooler. That means Summer is ending but more importantly signaling the arrival of Dragon Con, my annual pilgrimage to Atlanta, GA to visit friends and fight robots at the Robot Battles and MicroBattles Contests. Last year I brought DDT and Dead Meat (formerly Turboencabulator) but with Dead Meat, well, dead I needed another entrant to take over the main event. Luckily I had built a mega version of my original beetleweight, Ron, and could think of many ways to make him worthy of the stage.

After Motorama
MegatRON was pretty sad offensively  after Moto. In its fight vs Shaka, the dustpan was wrecked. Although Adam helped reweld the pan for the rumble, the pan was once again split open by Hyperactive and . To make matter's worse the saw arm motor smoked when I was using it as a hammer. It had a pretty sad ride back home next to the nearly victorious Uberclocker. Maybe some of that luck will rub off on it next time?

Uh... a bit more than rub off.

The Love Robot
To satisfy the rules of Robot Battles, the robot had a tip speed limit for all spinning weapons. This meant that the saw cutter could no longer be used with its current configuration. Instead I opted for a lifter and clamper mechanism similar to that of Uberclocker but utilizing the same hardpoints on MegatRON and general dustpan shape. After a first-pass concept, I flushed out the below design:

Looks too much like Uberclocker to me although the components are different. This design uses dual Banebots P60 132:1 gearmotors with RS550 motors. At ~10 inches from the pivot point, this configuration should have tons of lifting torque at the tip and more than enough speed to boot.

Additionally, I was feeling extremely lazy at the time. Not only was I taking courses and doing research but I had an industry short course of my own to teach and a conference the following weeks that needed a poster that I had not made yet. In short, I was strapped for time. If I didnt have the gearmotors on hand for the project, It couldnt be done.

This design turned out to be more feasible. Say, what is that beastly box object tied to the main lift pan?

It is a Tonegawa PS-050 super ultra mega cray servo! I found this sifting through old FIRST and decided to give it new life. This guy would make the top clamp arm at over 1200 oz-in of torque!

However, the servo has plastic gears which are still concern for stripping, especially in this high torque application where backdriving is a possibility. Similar to the lifting pan, I will integrate a slip clutch to relieve the servo under torque but also use spring steel strips as the gripper mechanism. My hope is that the flexibility of the steel is enough compliance to prevent damage to the servo.

The real challenge was converting the servo horn into a more useful medium. I had two options. The first involved milling a relief into the face of a shaft to match the servo horn geometry such that the horn could essentially sit in the relief and transfer torque by twisting against its features. The second option involved modifying the horn to essentially emulate these servo spine shafts from Servo City. I decided towards the latter for ease of support. Below, I describe the steps I used to create the part.

First I turned the horn into a round. No particular diameter, but the more nylon meat the better.

Next I counterbored both ends of the ceramic-coated aluminum shafting. One end should press fit the nylon servo spline but the other end diameter mattered less since that feature is only meant to pass the head of the servo horn screw. In this case, I needed a through diameter for the head of an M4 screw.

I pressed them together and drilled a #2 radial set screw for good measure. I could now assemble the servo shaft assembly.

After machining a few thousand standoffs, the weapon comes together with some long tie rods.

The top clamp servo is also mounted with standoffs. Now I can add the clamp clutch and clamp arm!

Pretend-o-bot! Looking good so far!

Very little would change with regards to electronics. I removed one of the 85A bot bitz escs since a single esc could power both lifter motors. To connect them together I utilized what I call the "shotgun connector".

I mean a double barrel bullet connector could be a shotgun right? Id love to see someone scale this into a six-shooter or something...

The servo is beastly enough that it would need far greater current than the 500mA or so from the ESC BEC. However, with a 12V max limit, I would need a method of stepping down the main battery voltage. Luckily, Charles had experiemented with DC-DC converters for herpy nano bike. He simply used a BEC and changed the feedback resistor. This tricks the converter into supplying higher voltage but at the expense of current.

After those changes it was time for a test! Seems to pass.

The robot wheelies very hard on accelerating which makes scooping robots or object very difficult. However should it happen the pan turns out to be incredibly quick! It is a standalone flipper!

Later that evening, Charles finished up his modifications to Uberclocker so we decided to spar.

From the video, you can easily see how detrimental the wheelies are. Most times, I jump right into his grabber. However, should I keep mobile the robot is still very maneuverable and has lots of flipping potential.

Aside from the expected scuffs, the robot took no damage and performed as expected. From this point on, I decided to tidy up the wiring and take the finish shots.

And this concludes the build report for.... what am I naming this thing again!?

Charles and I agreed the robot looked like a bug, with the red grippers being eyes or the spring steel strips being antennae. This immediately turned towards one of our favorite inside jokes, "bug love tree". This phrase was uttered by a professor during a robotics competition using broken English and has been cherished by us ever since only to rival that of "control systems potato". Change tree into robot and ship it. "Bug Love Robot", perfect.

How much love with Bug Love Robot give at Dragon Con? Read the event report to find out!

Friday, April 18, 2014

Debonair Jayce Cosplay v2: Transformation Complete

Last post I detailed the build for my Debonair Jayce Mercury Hammer prop. It was intended to do awesome things like transform and light up but failed because of my cheapness in design. Since Anime Boston, I've beasted through a tough 3 weeks (and $$$) to bring version 2 of the Mercury Weapon to Pax East 2014!

What Didn't Work: In a Nutshell

  • Dual purpose motors: I was being stingy with my spending and tried to make a component ill suited for the application work in my design. The double purpose motors were an interesting idea that should have saved weight and space but there are better alternatives.
  • Heavy metal \m/ (and other materials): I used materials that were free for me (1/4" 6061 aircraft grade aluminum for example) but absolutely overkill for a prop. They were durable sure, but EXTREMELY heavy compared to the pinkfoam standard of most props.
  • Color Scheme: The ivory was just a bad idea. Without a good primer behind it (plus recoating over old colors) essentially made it look like an old Macintosh computer. The static colors of yellow-gold placed adjacent were a nice barf color, esp when juxtaposed with the ivory. 

What Will Change: Minor Modifications

  1. Functional Mechanics: This means subdividing the mechanics to more "comfortable" technology, even if it costs me a bit more. This meaning I'll default to RC servos for the wings and try to make a screw-based mechanism for the rods.
  2. Actually Integrate Electronics: The previous version did not require an microcontroller because it was all DC driven at main voltage with clutches to compensate for physical stops. This version uses servos which input a waveform with particular properties. I would need a microcontroller or at least a 555 timer circuit for this. Although I am aware I could achieve this using passive components and a 555, I enjoy the visual interface of the MCU IDE and ease of change. The latter will be particularly useful for debugging values later (LED color, servo position, etc).
  3. Change of Materials: If it can be helped, all metal parts will be changed to high impact and tensile strength plastics. This should provide high strength at less weight and more color options! I would also like to try using this very popular pinkfoam that friends like Pizoobie use.
  4. Change of Dimensions: A second look at the hammer revealed to me that some of the proportions were off. I will attempt to remedy those effects when I update my components and functions
  5. Change in Color Palette: Although I still understand the weapon is majority silver/chrome, I still feel that the majority light with chrome detailing better suits the cloth aspect. Think of the weapon as a reflection of the user. With that said, I am only planning on simplifying the colors. Flat white (glossy or matte IDK) with chrome detailing. If you follow the skin in-game, the silver parts will be white, and the gold parts will be chrome. I will use LEDs this time for the areas that actually change between forms.

The Rebuild

Few hours with the SolidWorks models and the hammer was born.

The first thing I really did was think of ways to reduce the size of 3D printed parts. In a sense it is cheating with sculpting but the properties of honeycomb ABS are perfect for my goals. Still, I wanted to handle some of the load on my printers instead of just the IDC Dimension 1200es so I split the parts into sections that minimized supporting material.

The cut model shows a lot of the new magic inside the device. Instead of having a huge metal ovular plate, I split that section into halves. A skeleton frame of delrin plastic supports these floating panels and gives the appearance of a full volume. 

The purple rectangles are Turnigy brand RC servos. These will drive a 4-bar mechanism  that translates and rotates the wings outward. The pictures below elaborate on the new motion. The T-shaped bar is the piece where the wings will mount. The T shape in particular is to distribute the mounting points so it is more stable.

This is far more accurate than the single point rotation of the previous design. Furthermore, the 4 bar contains singularities which help reduce the constant loading on the servos. This will aid in their ability to hold position, a huge problem that plagued the first version.

What are those four cylindrical things? Glad you asked! I originally wanted to attempt a nested lead screw concept to extend the rods but I soon realized it was not feasible when I tried to design it. The second alternative are pneumatic cylinders. These provide quick response articulation of high volume linear motion in a compact initial configuration. Since nobody sells 2-stage air cylinders, I had to design them myself.

Air will enter through the bottom and fill the chambers composed of 3D printed endcaps and polycarbonate tubing. When pressure is released, an internal spring will retract the pistons. This is how I expected them to operate anyway. 

If you have pneumatics, you need to have air storage. Luckily, since I decided to move the batteries up to the hammer end, the bulb was now vacant. I stored a series air tank in the bulb and length of handle and left ports top and bottom to be filled and allow passage of compressed air.

Then the construction started. By this point in time, it was the week before Anime Boston. Could I finish it? There was only one way to find out! While I had the 3D printers running, I had the white polycarb and delrin waterjet cut at the MIT Hobby Shop.

Fresh prints of MIT. Three of these actually failed because I ran out of material qq.

Stratasys ABS P430 works in the Up Mini's! Slightly different color but much better mechanical properties.

The adjoining pod paneling were held together securely using a T-nut strip idea. No locating holes! just bore, tap, and arbor press!

Began assembling the oval structure in the center. Note how the ovals are cut in half. This eases assembly, servicing, and makes material conservation more effective. Also, that chrome!

Spray painted over the 3D printed ABS. Despite sanding, the striations are still very visible. At this point, I hadn't learned how important primers were yet. I opted for another coat to attempt a reflective chrome finish.

The bulb also suffered from that effect. Hilariously enough, I applied primer after the silver coat to get a clean flat white. I should have primed the entire structure, then detailed with the chrome. This probably would have helped with the paint bleed I experienced as the dripping paint follows the 3D printed layers like channels.

The chrome looks like wrought metal doesnt it? Actually kind of cool if I was trying to get that effect.
Now we get mechanical! Here is the compound air tank that will be storing my pressured air system. Its not a lot of volume, but I should be able to get one or two pops off for a good show.

This tank seats into the bulb like so. The mast goes over the longer storage tank.

Now we turn to to the wings. Cutting huge sections of double layer pink foam on the hot wire cutter...

... I can then hot glue to the table of the shop bot. What is a shop bot? An enormous CNC router! See the video below.

We end up with this.

And then this!

Because I was using a 1/2" ball endmill, there are some pretty heavy mill marks on the surface. Luckily, pinkfoam takes very well to sanding. I used an 80 grit sanding sponge to clean up the surfaces.

Experimental coating idea! My buddy Charles Guan made too much microballoon epoxy, so we slathered some generously over a spare block of foam. The result was a perfectly smooth hardcoat! Takes well to painting too! Below is a single coat of chrome.

It was so good, I applied it to my final pieces as well.

While that dried, I turned my attention to the electronics. I build a simple test circuit using a solderless breadboard, an arduino nano, one section of RBG LEDs, a limit switch, etc.

Since I was expecting using ~12v for the LEDs, I knew I would have to stepdown the voltage for the servos. A Turnigy voltage regulator drops the voltage to 6V instead of 12V. Test video below!

Once it proved to work, I moved it onto a solderboard. To be honest, leaving it on a solderless breadboard would have been more convenient. However, I see some potential in making cosplay electronics boards. hmmmm.

The board accepts seven inputs of RGB LED strips, and four servo inputs. To trigger transformation, it reads a digital input (like a switch).

Next soldering LEDs! If you were ever curious, you could actually achieve a variety of colors just by changing the voltage; you dont need to buy the LED strips with the controller. See this PDF for the details! Luckily, since I only needed yellow and cyan, I just needed to change which pins received full voltage.

Time to separate it from its power supply umbilical cord! The batteries were made from old and unwanted A123 cells. I wanted high capacity mainly (2300mah) but the added benefit of free and "beast mode charging capable". By the latter, I mean the lithium iron phosphate technology can be charged safety at 15 times its capacity, further meaning I can recover from low voltage at conventions quicker.

Wires and tubing are transmitted through the hollow polycarbonate tubing with a few holes.

With a bunch of sections done, it was time to simply put them all together. I was able to obtain some pneumatics components from old FIRST robots but had to buy a smaller gauge and cow-shaped manifold for the outlets. I was originally considering a network of tee fittings but it would have been too space consuming.

This will be mounted inside the side compartments with some screws.

Electronics mounted. The batteries were shrink wrapped to add protection. They are velcro mounted to the sides.

Remember those four bar linkages? Here they are fresh off the waterjet! Cut from 1/4" 6061 aluminum. The purple pieces are aluminum servo horns from Turnigy.

Installed on the sides.

Here is an important less about servos: not all servos are made alike! I had them wired previously to share two digital pins. Technically this should have worked since the servos are opposite each other with respect to the diagonal plane. However, we find they are not alike and need to be individually addressed in order to fine tune position. I ended up taking two of the signal wires from the servos and soldering directly to two digital pins on the arduino. A bit dirty, but should work.

Updating code...

To attach the foam pieces, I was going to hot glue weldnuts on the backside. This way I have something secure to bolt into and allows for the pieces to be removable. To make the holes I simply aligned the back pieces to the foam and poke holes into the foam. Then I bore it out by hand to a final diameter and lightly counterbore with a 3/4" endmill.

All four!

Wither those attached, I could do a semi-real simulation! Here is the hammer...

... and the cannon!

And video!

Now for the last two days, it is time to rush build the pneumatic pistons. How much have I gotten done so far? Well uh...

2 hammer Jayce <3 br="" nbsp="">


I didnt get to go until Saturday and even then I woke up late! I rushed hurriedly to the Boston Convention Center in hopes of making the cosplay show in time. I didnt even enter the doors before I was quickly noticed. After the first transform the reaction was insane!

I met some cool guys...

And ended Saturday with this nice featurette on the League of Legends tumblr. 

Just a four second video that was reposted on reddit.

And I became semi-famous instantly. At least for a day anyway. :P


Reflections and Conclusions

This weekend has taught me that whatever I cosplay, I MUST make it mechanically articulated and electronically fancy! First, I kinda enjoy the attention (even if it does take me 30 mins to go down the hall) but more importantly there is some great educational opportunity here. I would say  70% of the people who took my picture wanted to know how it worked and while that is an incredibly broad question I was able to stimulate good discussion about some engineering topics.

I believe cosplay can be the intersection between art and engineering. If I can build something to capture an audience, I can hopefully inspire them to investigate and learn more. This goes for anyone, no matter how young or old.

Otherwise, I really need to start these projects sooner. 1 month is an impressive number to put up but boy am I exhausted qq. I'll be spending now till Dragon Con fixing the pneumatic pistons, which should be a manageable task if I just keep up this pace.

Useful Links

Arduino microcontrollers
Hobbyking: great source for batteries, RC equipment, and Arduinos
Big Blue Saw

Special Thanks

  • Thanks to everyone who was a part of the making or picture taking of this version of the hammer!
  • Charles Guan for hosting a build space, helping with printing, mixing too much epoxy, and helping with machining!
  • James Penn for taking the semi finished pictures and video!
  • Everyone for letting me know that I got fed on reddit!
  • Riot Games for making an awesome champion and posting that mini video!
  • Cathy Heo for your moral support and pro sanding ability!
  • Brian "Iron Man" Chan for use of the Hobby Shop Waterjet!
  • Amy "Pizoobie" Zhao for her consultation on foam coatings!
  • Cynthia Lu for finding the RGB LED guide!
  • Thanks to my PI Anastasios John Hart for letting me slack off a bit that week XD
  • Happy Birthday to myself!
  • And last but not least, thanks to the League of Legends Cosplayers Facebook group for being a constructive and awesome community! I think we can all agree the positive atmosphere in that group keeps us coming back to make another costume!