Big Data: Test & Refresh

When I built the first Big Data in 2013, I was just get­ting to grips with my new sur­round­ings. I even added the EVT moniker from my new job: in hard­ware devel­op­ment, “engi­neer­ing val­i­da­tion test” means both test­ing the ear­li­est pro­to­types and the pro­to­types themselves.

So overweight it needed its letters clipped...


I was over­whelmed by the Bay Area and toned down my typ­i­cal­ly masochis­tic design phi­los­o­phy. I used a less-exper­i­men­tal (proven, even) weapon sys­tem1 and relied on tried-and-true water­jet-style design.

I allowed one indul­gence, which was build­ing a new brush­less con­troller appro­pri­ate for this scale and for the sen­sored 22 mm trac­tion motors.

This is all great except the mechan­i­cal design did­n’t work. The wheels were nowhere near the cen­ter of grav­i­ty, so that it did­n’t real­ly move except in short bursts, and the trac­tion motors were each held in with five threads worth of four M2 screws on the face, which is to say they weren’t held in.


Then came Big Data DVT (design val­i­da­tion test) and the start of the orange. Embold­ened by famil­iar­i­ty with new capa­bil­i­ties, I used the CNC mills avail­able to me and cast­ed parts from polyurethane. Trac­tion motors got even beefi­er while the weapon became more opti­mized… to the point of unre­li­a­bil­i­ty. The wheels shift so close to the cen­ter of grav­i­ty that just 0.54 g of accel­er­a­tion caus­es DVT to pop a wheelie.

Naked wheels held on with un-thread­lock­ered shoul­der screws reflect­ed an opti­mism and non­cha­lance to robot build­ing of col­lege days past. The trip to Atlanta, comped by asso­ci­a­tion to Google’s recruit­ing at Geor­gia Tech and house calls to the local dat­a­cen­ter, laid excru­ci­at­ing­ly bare how not to con­struct a vaca­tion (or robot).

An aside: there’s a cer­tain art to express­ing your­self dur­ing a bot match. Jamo is a vir­tu­oso both with the R/C trans­mit­ter and with sit­u­a­tion­al­ly appro­pri­ate utter­ances of raw emo­tion. I might be on the oth­er end of that spec­trum with my twitchy fin­gers and a vocab­u­lary of only “oh” and “no.”

Big Data DVT CAD

Also of note is that I’ve ditched machined hor­i­zon­tal chas­sis ele­ments com­plete­ly; i.e. there is no top or bot­tom plate that holds oth­er com­po­nents in place. I’ve favored this ver­ti­cal con­struc­tion style since build­ing sumo in high school. I reused the round-stand­off-between-plates tech­nique used in Flap­jack­’s base because of the easy assem­bly and smooth curves it lends to a design. In Flap­jack they were used to mount bear­ings for the shell; here, one of the beams—a hard­ened steel shaft, bored through and thread­ed on both ends—likewise sup­ports the weapon from the milled alu­minum “book­ends.” To not load the screws in shear, the chas­sis is coun­ter­bored for the shaft to nes­tle into.


With all the cool log­ic of a frus­trat­ed tod­dler spurned at the toy store (“I WANT IT”), Big Data—and by Big Data I mean I—doubled down on the design out in the park­ing lot (“BUT BUT I NEEEED IT”). Pro­duc­tion ver­i­fi­ca­tion test (PVT) rang me up demand­ing none oth­er than a weapon and dri­ve­train each tak­ing up 30% of the weight. That left just 40% for a weapon motor, chas­sis (with all the bits & bobs), and electronics.

Big Data PVT

Where­as DVT used alu­minum trac­tion motor mounts fixed to the chas­sis’s two round beams, PVT’s dri­ve motors were con­tained along with PCBs and bat­ter­ies in a 3D print­ed “scaf­fold” wrapped around its three frame beams.

Machining right chassis case
Test fit right chassis case
Test fit whole chassis

PVT was the first Big Data that entered com­bat with any sem­blance of armor. Its out­side alu­minum rails (15 g) served as dou­ble sup­ports for the geartrain dead shafts and pro­tect­ed the wheels and gears a bit. They’re pretty.

Big Data PVT side rails

Unfor­tu­nate­ly, my cus­tom motor dri­vers were less pro­tect­ed against the ESD haz­ards of dry win­ter air. Nor could the 3D print­ed scaf­fold with­stand direct attacks from Chad’s saw and Kyle’s hor­i­zon­tal bar.

Assembled Big Data PVT

Why do I have to fight Geor­gia Tech RoboJackets?



PVT was under­weight by at least 80 grams, and the frame shows basi­cal­ly no dam­age. Its access hatch made it much eas­i­er to work on than any oth­er Big Data. If it weren’t for the too-wimpy weapon motor and that I mis­placed half the chas­sis, I’d even con­sid­er using it again.


I sold out. Brush­less trac­tion pow­er affords me ridicu­lous speed but I’m not a good enough dri­ver nor do I have enough time to prac­tice to take advan­tage of that. And to be hon­est, I was get­ting a tad bored with this design.

Big Data 4

I acqui­esced to the voic­es call­ing for a big­ger weapon motor and small­er brushed trac­tion motors. Big Data 4 final­ly has armor, and loads of it. I even tossed in my first ever wedges in combat.

Hoop armor wedgelet

Although at Drag­onCon 2015 I kind of just knifed those in.

Whittling UHMW hoop

Fear of Design Spaces

Anoth­er aside: I suf­fer from what I call cre­ative ago­ra­pho­bia. I’m par­a­lyzed by too open of a design space, too many free vari­ables, too few con­straints. When I’m design­ing, this is the screen that pan­ics me the most.

SolidWorks user interface

Hav­ing a free-to-use weapon bar and a gung-ho atti­tude to brush­less trac­tion were “hard” con­straints that helped shape the last three Big Datas. I’ve heard that called “com­po­nent-based design.”

Con­verse­ly if I con­sid­er the frame design a fixed estab­lish­ment, trans­formed from free vari­able to set con­straint, it gets eas­i­er to tweak oth­er com­po­nents I took for grant­ed, like the sil­ly-ly large trac­tion motors. To get a design that worked, it was time to back off from those Portescap motors.

Maxon 139885 Gearmotor

These are 13 mm diam­e­ter Max­on 67:1 gear­mo­tors, sold as “235 RPM” units on eBay. They’re… OK. The con­struc­tion is great, with large robust steel gears and gear­head mat­ing threads built into the motor itself. But, they’re def­i­nite­ly slow for beetleweight com­bat and their pre­cious met­al brush­es do not enjoy over­volt­age oper­a­tion at all. Maybe they become more use­ful after remov­ing a stage of plan­e­tary gear reduction.

For now, they serve to pro­vide slow but torquey dri­ve trac­tion for Big Data 4.

In its debut fight, BD4 prompt­ly gets bot­tomed out like a Google shut­tle at 23rd St.

Google Shuttle Stuck

At this point, the hoop armor has yet any wedgelets at the front, although they’d be inef­fec­tive any­ways against the floor-scrap­ing scoop on Wedgee. Also at this point, I’m still tip­sy from the night before and have had three hours of like­wise inef­fec­tive sleep. Which some­how leads to bet­ter dri­ving than ever before. ¯\_(ツ)_/¯

Of course in my next match, I dri­ve myself right out through the pushout.

Two months lat­er I fight Aaron’s Mar­gin of Safe­ty, the new Drag­onCon cham­pi­on. It went as you might expect.

So before I get to the design vul­ner­a­bil­i­ties this revealed, let me get to the crip­pling mis­man­aged state that BD4 was in for the Com­Bugs Cup fight. None of the screws had thread­lock­er on them because I had tak­en Big Data apart for ship­ping. The loose left-side weapon shaft screw comes off with a “PING!” and a black blur at 0:39 in the video, so the shaft is sin­gle-sup­port­ed after that.

Big Data 4 Eating Itself

Also, the set screw that holds the weapon motor in its fire­wall mount had nev­er been tight­ened. This lets the the motor slide out of posi­tion and… into the weapon bar.

Big Data 4 Weapon Motor/Pulley

On the oth­er hand, the UHMW hoop armor works well at keep­ing Mar­gin­’s disk away from the vul­ner­a­ble nylon shell. It’s not low or thick enough to pro­tect the wheels, but that’s a con­stant dan­ger in any fight against an undercutter.

Big Data 4 Post-Fight

The killing blow (1:36 in the video) was an incred­i­bly vicious hit with Mar­gin­’s sin­gle tooth engag­ing the full wall thick­ness of the chas­sis piece.

Big Data 4 Frontal Damage

It applied forces towards the out­side of the piece (i.e. towards the hogout wall). This con­cen­trates stress into the sharp inside cor­ners of the hogout, caus­ing a frac­ture that prop­a­gat­ed into the walls of the bulkhead.

Big Data 4 Chassis Damage

As designed, the piece is meant to break or wear down at the low­er cor­ner when hit with­out dis­tort­ing the frame geom­e­try. Unfor­tu­nate­ly it frac­tured way more than expect­ed and the loss of the weapon shaft screw allowed the shaft coun­ter­bore to dis­tort, mak­ing reassem­bly impossible.

Big Data 4 Dead Piece

How­ev­er, the “hard parts” like weapon shaft and wheel axles sur­vived with min­i­mal dis­tor­tion or thread dam­age, ready for reuse in anoth­er frame. All the motors and dri­ve­train are fine as well. For my first fight with a spin­ner, it was­n’t alto­geth­er that bad.

It’s an engi­neer’s nat­ur­al instinct to tin­ker with an exist­ing design. How­ev­er, I’m not going to mess with Big Data too much; I promise to refine and ship a more reliable—not sexier—product.

See you at Motora­ma 2016?

  1. Although that would be a first for me, so does it still count? []