The media cart of science, part deux

The media cart of science proves itself ever useful once again:

The Sambo

Oh yeah, and Team Cornell now refers to sandwiches as Sambos for some reason. In case you were wondering, sambo rhymes with Rambo.

Don't ask me, because I really REALLY don't know.

Bedknobs and bumper replacements

So... a LOT of things happened last week. Of the five weekdays that went by, I think I ended up eating dinner after midnight for no less than three of them. We had three guests from Singapore, Dr. Koh, Jason, and Hikaru. This is what they look like after a multi-day flight over the North Pole:

Jason and Hikaru will be coming back to work with us, as soon as they get student visas.

We also had demos from several sensors companies. Anthony from Velodyne brought over a fancy 2D LIDAR. I say fancy because the 1D LIDARs that we used last year only scan along one line of laser every time they record. The Velodyne scans 32 lines at each data step, and even has better resolution in each line than the 1D sensors. Anthony brought a demo unit, so we gave it the "Noah Special" (i.e. we whipped up a sketchy last-minute mount to our test vehicle):

Despite weighing 40 lbs. or so, it did actually stay on the roof of the Suburban. A good thing too, because it would've been a very costly $75K mistake if we'd dropped it. If you look carefully at that picture, you can also see a fancy 1D LIDAR + camera mount that Pete conjured up for the Suburban's front bumper. Those sensors were originally supposed to be mounted to a "Ranch Hand," a delightfully overbuilt bumper replacement for the Suburban (used mostly for clearing cattle out of the way). Unfortunately the "Ranch Hand" manufacturers discovered a slight wrinkle in the plan- they don't produce it anymore. This startling discovery came after a disturbingly long series of events, which included money changing hands and several calls to dealers and the manufacturer for the aforementioned "Ranch Hand."

Oh yeah, and we got some vehicles. I'm not exactly sure why that news got lost in the shuffle, so my only excuse is that we really were that busy. Anyway, they are a '99 Suburban for testing and general demolishing, and a brand new 2007 Tahoe for the competition:

Adam's song

Adam and Max, two of our new recruits, have been working on a slick networking trick with our LIDARs. They want to make each LIDAR its own separate entity on the vehicle network, so it doesn't have to pass data through a computer. Why? Well, last year we needed a separate serial card for each LIDAR, and we actually ran low on PCI slots in our computers. It worked fine for the 3 LIDARs we used last year, err, so long as the computers didn't crash (in other words, it didn't work so well).

Anyway, this year Adam and Max are making each LIDAR communicate directly with the vehicle network through its own reader microcontroller. The micro would act like a liason between its sensor and the vehicle network, so each sensor would be independent of the computers. It apparently also allows for some killer pulse-based timing, so we bypass sketchy Windows thread scheduling altogether (imagine trying to time-stamp something when you don't know when it was created).

They have the system working, err, mostly:

One of the weird things about microcontrollers, however, is that you never quite know what's going on inside them. They operate at a much lower level than something like Windows, so they can be a pain to debug. Sometimes even getting the silly little red light to flash is an all-day accomplishment. This song, which can be sung to the notes of Blink 182's Adam's Song, describes their plight:

~I never thought~

~this wouldn't work.~

~The HCS12 is being a jerk.~

~I traced the LIDAR to its card.~

~It really shouldn't be this hard.~

What, you want me to finish the song? Psh, I have work to do; nevermind the fact that it took me all morning to come up with it.

DARPA Urban Helpdesk

"DARPA Urban Helpdesk, how can we be of assistance?"

"Any of you guys know SolidWorks?"
"Oh, you want Noah, at desk 2."

"I have a small squirrel problem."
"Pete?"
"Stomp 'im."

"Can you give this prospective student a quick explanation of CUSAT?"
"FOR THE EIGHTIETH TIME, WE ARE NOT CUSAT!"

The media cart of science

Sometimes (where sometimes means most of the time) putting stuff on a car to test it can be far more trouble than it's worth. The problem is so tough that often it's better to try to get stuff working in simpler situations before you send it to bat. We've been doing a lot of "simple" tests like that recently, where we lug our sensors outside, get some data, and spend some time processing it in the lab. Considering last year, where getting the car out was actually more of a quest than anything else, it's probably a good move.

Anyway, when we lug our sensors out, we need somewhere to put them. It's a job filled masterfully by... THE MEDIA CART OF SCIENCE [echo] Science science (science). Here are a few of the many faces of this versatile device:

Yes, that is a battery powering the computer, through a 1500 Watt inverter. Overkill? Us? Never.

Oh, and I thought I'd leave you with a quote from the lab:

"I got it working at 40 Hz."
"Sweet. Did you fix the bugs yet?"
"Yes... err, sorta. Well, not really. No."

If I had a million dollars

If I had a million dollars
~if I had a million dollars~
I'd buy you a robotic car
~a nice reliant automobile~
If I had a million dollars
~if I had a million dollars~
I'd buy you sensors for your car
~maybe a nice 2D LIDAR and some GPS~
If I had a million dollars
~if I had a million dollars~

DARPA has just announced that Team Cornell is one of eleven DARPA Urban Challenge teams to be awarded a research contract of up to one million dollars in support of our efforts in the DUC. A full sixty teams that applied for such funding, so we feel very honored and fortunate to be chosen as recipients. A more official (less candid) press release is available on DARPA's website:

http://www.darpa.mil/grandchallenge/docs/Track_A_Teams_Announcement.pdf

In other news, DARPA has apparently changed the text of the DUC rules so that first, second, and third place finishers in the DUC receive trophies. What does that mean? It means that Congress may very well deny DARPA the authority to give cash awards for any cool research that gets developed this year. Way to go, Congress, way to go.

Actually, considering that most of us university kids are here for the thrill of the chase, I'd say it's fairly shrewd of them. Then again, maybe I give them too much credit.

Anyway, funding changes everything. EVERYTHING. So here, fresh from the mind of everybody's favorite author, are the top ten things CUDGC won't have to do this year:

10. Mow lawns and perform other landscaping tasks to make rent.
9. Use cardboard boxes in place of real sensors and computers.
8. Invert matrices or solve ODE's for spare change.
7. Blind-call companies for a nut, bolt, or spare connector.
6. Deficit spend against the department's credit.
5. "Borrow" internetz from certain nameless people.
4. Rack up thousands of dollars in parking tickets for shady unmarked U-Hauls and half-built rock crawlers.
3. Buy malfunctioning radar units from sketchy dudes who won't accept return shipments until threatened.
2. Shift gears on certain cars with a bit of twine and a stick.
1. Replace the transmission every 3000 km.

The sad part is that a good number of these weren't just made up for the sake of comedy.