Project Seedling Comes to an End

At the end of the semester, there will be games.

-me, a few months ago

And games there were. One of the most largest-scale, most time-consuming projects I’ve ever set out on has been quietly taking place behind the scenes, mostly unblogged and unpublicized. That project was the 2.007 Electric Vehicle special section. Recall, again, the ‘billboard’ from the beginning of the term:

I’m proud to say it essentially went off like that – minus the Segways, which I’m kind of glad did not happen anyway. The last update from my own “undergraduate victory garden” was in March, so whats happened since then?

By the end of March, all of the vehicles were pretty much mechanically together. The big “rolling frame” inspection and checkoff was at this point, and the week before was filled with a flurry of activity and skipped classes.

Above, the “Skull Skates”, presumably named because they would easily crack your skull on the road if given the chance, get power tested for the first time. It took quite a few design revisions to stuff the brick-sized battery, outrunner, and Hobbyking cartroller into skate form.

The 2-person go-kart effort was well on its way to completion at this point. It was christened “Melonkart” due to the combined budget of the team being enough to afford the C80/100 class “Melon” motor and a big enough Kelly Controller (the KBS48121 series) to feed it, and because our idiosyncratic naming method has rubbed off on the undergrads.

Team Melonkart has individual builders’ websites here and here.

Another one of the “leader pack” of students who worked quickly early on is this scooter of a very familiar construction technique (ahem….). Its innovation is the large rear fender brake, which isn’t a Razor style fender brake so much as an elaborate brake lever which actuates a stock scooter drum brake. The idea was to prevent excessive tire wear from braking.

“Crüscooter” has an excellent writeup at its own website.

The one-man go-kart ran into some budgeting problems, so appropriate cost cutting measures were made. This thing was massive, smooth, and handled either like a Cadillac or a battleship, depending on which student you asked. The joinery on the wood had some particularly detailed work put into it.

Around the 3/4 mark of term, I ran to Radioshack and practically cleaned them out of small switches and buttons – people needed logic power switches, lighting, precharge switches, etc. Some interesting “switch panels” came about as a result, with the wooden caddy-ship (called “Splinter Cell” officially) having a particularly nice panel complete with serial port access for the Kelly controller. Above, this student just carved the panel-mount switch holes directly into the sides of his aluminum U-channel. His design also had a  “lid” that also acted as a deck, meaning the insides were easily reachable for servicing. That came in very handy at the end during the debugging rush.

The above scooter, “Railscooter”, named due to an unfortunate encounter with the railroad tracks that divide MIT’s campus into north and south, has a writeup by its builder.

And then there is this. Overall one of the favorite vehicles of this term for everyone, a $40 Craigslisted red tricycle was converted into a 24 volt electric fun/death machine. The project was elegant in its simplicity – what mechanical work there was to be done was performed exactly as described and designed, and on time. The back of the original vehicle made a convenient electrical mounting deck, so minimal new attachments were necessary, and the wiring was very clean. If I had to pick a “datum” to Pugh Chart everyone this term, this would probably be it. And it’s also awesome. Officially named “Travis”, it was a big hit at the 2.007 main final competition with the audience.

More from the Unlimited Class (not really, just not a scooter or go-kart) is the electric tri-board. Due to the flat nature of this vehicle, a custom battery and charger solution was specified. Conveniently, the battery came from a disassembled hybrid vehicle pack that the Electric Vehicle Team had sitting around – it used 4 4-cell sticks of D sized NiMh cells for a 19.2v & 4.5Ah electrical system, like an A123 brick and a half. The shell on the bottom is custom vacuum formed.

While it hasn’t been updated yet, the triboard’s designer has her own website.

competition day

The competition this year had two components: a 50 meter drag race, which was fairly simple, and a time*energy “hillclimb” race up a 4-storey parking garage. YES, that thing we keep doing in the videos of everything has now been made an official practice. That took alot of explaining, by the way.

The morning of competition day, people making last minute fixes and changes… Even when the vehicle was technically due the week before, there’s always things to mess with!

The drag race was held in a relatively smoothly paved back alley under the Brain & Cognitive Sciences complex. The track was 50 meters in length, demarcated with white tape. We sort of went all-out hardcore with this event setup, getting 2-way radios for fast communication and even glowy traffic police wands to start people with.

Scooters averaged 9 to 11 seconds, and Melonkart hit an 8.28 second run – with all the vehicles having similar power levels, we weren’t expecting blazing times, really. Of course, the instructors weren’t just there to stand by and watch – Tinykart hit a 6 second run (due to its 100% more motor power than most of the class), and Chibikart somehow also managed an 8.26 second run.

We started becoming short-staffed in the garage and the lighting was much worse, so there are relatively few good pictures of people running up. The garage hillclimb race was a hell of alot of work to set up, but it was the most entertaining in the end, and it’s something which I haven’t seen done before, so I think it was worthwhile.

The biggest time consumer in setting up was appeasing the safety office. First of all, I would like to point out that any safety office which doesn’t bat an eye at “We want to close an entire MIT parking garage and race student-built go-karts up it” is pretty awesome. They only requested after inspection of the plans that we have some method of preventing people from faceplanting at 20mph into the narrow turn ends of the garage, which I found to be a reasonable request at any rate – while it is about 40 feet wide in the parking lanes, it narrows down to less than 20 feet at the turning ends, so if you miss an apex or lose traction, there is a concrete wall waiting.

We dismissed the idea of using a wall of haybales immediately – because it would have taken a literal semi trailer of haybales to line the problematic areas. Instead, we came up with a method of tensioning construction debris netting between steel cables and looping the cables around the outer guardrail of the garage itself. The cables were tensioned with large turnbuckles. In the end, the result was a bouncy net of happiness and totally not decapitation I swear to Robot Jesus guys.

Shane and I, along with some of the students from the section, spent about 15 hours on the Saturday before and about 2 hours the day of the event making and setting these suckers up. There were 7 of them total. By the end, we were pretty fast at it – one could come up in about 10 minutes and back down in roughly the same time.

The cool part about it is, now that there exists an official activity safety process for this kind of event, we could throw a go-kart race almost whenever ……we feel like setting those up again. I was pretty amazed at the willingness of MIT Parking and the Campus Police to just turn off an entire parking facility (Granted it was on a Sunday when there were like 10 cars ever anyway) so we could race things in it. Legitimately.


What’s better than building a vehicle to troll confused campus tourists with? Having MIT pay you to teach others how to do the same. Good job to everybody, and thanks for the excellent “plaque” at the end:

Finally, my own thoughts and reflections on this semester.

The rules of the game were pretty simple. You had to use 1 to 3 of the A123 12V7 bricks in your design, or else if you do want a custom battery solution a charger must be included in the budget. You got one 8″ pneumatic tire for free, choice between one with a sprocket, one with a belt pulley, and a “front” wheel i.e. no  attached drive parts. You didn’t have to use it – this was a last minute pre-term rule change, because I was about to make everyone use an 8″ drive wheel. This was to encourage some more diversity in design (a point i will address later), and in the end, I’m glad it happened. There were just some seriously creative efforts that would have been hampered by a wheel requirement.

Major components, including motor, controller, frame materials, power transmission components, and any other vital parts (such as the deck, for the only skateboard-style project) must be under $300 not including shipping costs. Hardware and some small incidental metal stock was not included – I kind of set $10 as an accounting quantum for purchases. A few people bought long rods or large plates and then split them between vehicles. Students aggregated their orders with me and they were sent out every Monday and Wednesday – if you didn’t make the 9PM cutoff on one day, you had to wait until the next, and you had to accept that whatever you order (that wasn’t from McMaster…) was probably going to get here several days from now, so your order had to be well-planned. Overall, with all the overhead of shipping and consumables (wiring, hardware, etc.) included, the net cost per student ended up being more like $500. I’m generally content with this budgetary wise – if I had to buy every part of one of my random vehicles from scratch, it would cost several hundred dollars, no question. The caveat, of course, was that nobody had to pay for batteries thanks to the generous donation by A123Systems of something like 24 lithium ion SLA-replacement modules (ALM12V7 type) which would have run like 4 figures if we had to buy them or a similar product on the market.

One thing that impressed me was the diversity of designs that came up. While most people opted for scooters of some sort, there was a longboard, a set of inline skates (which ran on 12 volts each, using 2 batteries total), and an electric drive conversion of a little red tricycle. Yes, that actually happened. Three students went for go-karts, a team of two and one independent. Encouraging this kind of design diversity and letting people explore the same fundamental concepts using their own methodology is something I strongly support and encourage. It’s how I learned practically all of my engineering and design knowledge, and how I know so many different resources and some little-practiced skills among college students such as how to order things on McMaster and Digikey. Quite a few people this term learned how to pick parts of desired specification from huge catalogs, I think, and that is a skill which will serve them well.

Yet at the same time that I praise the ability of several particularly motivated students to go above and beyond and create very ingenious designs, I’m aware of the limitations of academia and the availability of resources, even at a place like MIT (which I make seem like a bottomless well of free parts and machining at times, admittedly). While financial limtiations and parts availability was not a concern for us this year, instructor time was. The format of the class right now, in my opinion, is not reproducible or sustainable without the massive involvement of one or more instructors who know the field in and out – who knows, for instance, that Kelly KBS controllers are unreliable if run at full motor current, or that R/C controllers require some special design considerations in the drivetrain and cannot be started from standstill, or random little tricks you need to pull with the motors’ Hall sensors to find the optimal running timing.

While it seems like that should be the optimal model for teaching, the fact is that I literally cannot do this again – the amount of time this section took up this time means I’m behind on finishing my required graduate level courses. It sucks, since I really do want to run this section again, but it is also of little value to MIT or other universities (like SUTD, for instance) if I am the only person who can do it. You might be able to construe it as job security, but I’m also not that into academia. So whatever this ends up being, it should be maximally teachable by anyone – the course content and procedurals must be robust enough to stand on their own.

Ultimately I think some of the design freedom has to be removed. It’s difficult to keep track of who was using what parts and what was different about each design. What I haven’t settled on is how to trim down some of the extremities of the design space of small vehicles without constricting too much the students who want to take the class and run with it. One issue that also came up was the final competitions – different vehicles required different safety considerations and the final way that I worked out with the safety office here to secure the venue against go-karts was perhaps overkill for the scooters. One-size-fits-all works both ways, and while one kind of educational experience doesn’t work for everyone the best, so that one experience cannot cover all of everyone’s desires simultaneously.

I do have some thoughts though. First and foremost, what is definitely not sustainable is letting everyone keep their vehicles at the end. It’s great when a class has takeaways, but the more random vehicles (of questionable reliability and safety even if they passed mechanical inspection at the end of term) that are released onto campus, the more chances there are of something bad happening. Not all the vehicles are street legal either, like the go-karts, so the grand mission of letting people build their own transportation around campus cannot even be fulfilled there. Ideally, in this case, the students would take away what they learned in the class and should they want to, use that knowledge and immediately turn it around and build a self transportation implement using their own resources.

Second, the progression of this section in the past few years has already been a game of selectively removing degrees of freedom from the curriculum while maximizing the student engagement. It’s almost like designing design itself….which, interestingly enough, is what the professor/research group I currently work for is focused on. Couldn’t tell with this whole “running entire design class” thing, right? Last year (2011, when I was not directly involved in running the EV section), the students had almost 100% control over their parts selection, including wheels and batteries. This was a boon for the one or two people who actually could take advantage of it, and who knew where to find stuff to use, but the rest got lost quickly lost and frustrated. The result was that we had only 3 people out of 5 complete anything at all, and of those, only two were reasonably well-executed. I decided to passively remove the wheel choice (“You could use this, but you don’t have to”), and that seemed to help some students get started in the right direction this year. Batteries, too, were a limitation, though not really premeditated.

So what other degree of freedom could I remove which would let the class retain as much of its design freedom as possible but also let it be standardized and accessible to those with less experience? Pursuant to the desire that perhaps not everyone should be flying around on their vehicles after the class is done, I’m looking to the likes of Tinykart and Chibikart for inspiration. We’ve kind of dubbed it the “1-inch extrusion frame” class of racing, and it’s appealing me for a few reasons.

First, it goes together quickly while still letting you retain alot of design freedom. It’s not like the standard scooter tactic of using an aluminum U-channel as the frame, for example, where even though you can put together the frame quickly, it still looks like a u-channel. The whole point of 80/20, or regular metal tubing applied in a similar fashion, is the ability to create prototypes of machines or test rigs or kitchen tables, etc. by dictating how the extrusions connect. You would have to design and fabricate the joining members and also figure out how to attach other components to the 80/20. It’s also almost infinitely adjustable – don’t like where the motor is? Loosen the slot nuts and slide it over.

Second, it creates a bit more level playing field for students. In the competitions this year, we clearly could not run scooters vs. go-karts – the levels of power and speed each of those needed was clearly different…and then you throw in the trike, the longboard, and the roller skates. It’s hard to get a historical trend going when the vehicles are so different that they cannot be classed together, if at all. It would also have the effect of standardizing the parts and materials a little – instead of placing like 6 orders to SpeedyMetals for various sizes of aluminum channel and plates, you would start out with a certain amount of 80/20 and a few big plates of aluminum to cut whatever you want using any process appropriate.

Third, it would ease up on the instructor burden. We would have a better idea at the outset what is needed for the class – I wouldn’t be placing a different order to Hobbyking every week during the term. All of the vehicles would require much of the same level of work to debug. This would ultimately involve phasing out some of the more hackish solutions like R/C controllers, but in my opinion that is for the better from a reliability perspective. It lets the students have more time, too – instead of waiting a week for their HK order to come, for instance, if we understood that there is a selection of motors which is the most popular, then it can be ‘buffered’ beforehand and people can start immediately. That, along with more scheduled checkins and demonstration deliverables, should prevent even more people from falling behind (I’ll say that the problem this year was much, much less than past years).

So to end this now thesis-length post, here’s my idea for a “2.00gokart”, as affectionately named by the students:

  • Teams of 2 – this year, the one-person go-kart effort ran into alot of time commitment issues, since it IS alot more mechanical work, while the 2-person kart team demonstrated good work sharing and planning skills. A kart being a more complex system, I think it justifies two person groups. Often times, people can bounce ideas back and forth between eachother, a chance they would not get, or at least wouldn’t go and seek outm if they were working alone.
  • You get 3 6-foot sticks of 1″ 80/20 extrusion, and a 24×24 aluminum plate in both 1/8″ and 1/4″ thicknesses. So, the main structure of the vehicle is extrusion with the plates being for joists, motor mounts, etc.
  • You may select up to 2 8″  pneumatic tire in chain drive, belt drive, or no-drive for free, but you are not required to use them (other wheels must be bought in the budget). This means some amount of spec’ing and shopping has to be done for at least one other wheel unless the design is very….special.
  • You may use 1 to 3 A123 12V7 modules for batteries. Going to 4, in my opinion, opens up a very clear optimal solution since using 2 in series and 2 in parallel gives you the highest watt hour count onboard. That’s no fun.
  • You have a choice between several different motors, DC and brushless, of various makes and models and parameters. A range of controllers will also be available. Part of the lab assignments would be to pick a motor and controller and justify your drivetrain design, just as it is now.
  • A $150 budget for drivetrain parts, special hardware, etc. will be made available, and must be justified/recorded, as it is now. Basically, taking out the value of a good Kelly Controller and HK outrunner with a McMaster order of frame materials and scooter parts dealer order of wheels and tires, etc. from the $300 budget.
  • The vehicle must be “kart” style, sit down, or otherwise statically stable. I’m thinking right now of a “casterscooter” with two inline wheels, one driving, and two little casters on the sides so it doesn’t tip. This is a horrific idea, but is one way to abuse the rule for sure!
  • The frame may not exceed 30 x 36″ (just an idea for a size range – most of the karts are about this big)

I totally think this is doable. The only question is, will I have time to do it again?

This is now the longest post I’ve ever made on this site, I think. What am I forgetting?

Oh right, the video!

2 thoughts on “Project Seedling Comes to an End”

  1. Video came out excellent, and the garage runs were a blast.

    I was blown away by the ‘legit’ factor of the safety netting and moreover the whole event organization; even from afar it looked like a boatload of effort.

  2. Definitely looks like more fun than any class I had in college.

    As I recall, FIRST gives contestants a battery, motor controller, and motor that are all designed to work together. This cuts down on the number of “moving parts” so to speak.

Comments are closed.