Our Curruent Project is building a
hybrid Cobra muscle kit car

The first big deadline for the 2005 DARPA Grand Challenge was March 11^th. Every team had to get a video to DARPA headquarters showing their vehicle operating under autonomously (at least a little bit) and under remote control. DARPA will review the videos and schedule site visits with those teams it thinks have a chance.

One Month from Deadline

Finish wiring up all the motor FET boards and motor encoders to the PCI Motor Controller board (have had problems with motor controller software, which setback some of the circuit testing).  Begin initial testing. We have one man on fire extinguisher, one man on kill switch, and one man on Camcorder.  First wheel works perfectly. Second wheel goes forward wel l, then reverses well for 2 seconds, then FIRE (#1).

Diagnose problem: The over-voltage circuit kicked in protecting pre-amp board and PCI Card. Believe cause to be an off-balance wheel that generated an above-tolerance voltage spike across the batteries.

<Pictures are not in sync with story>

Smoke

$120

Three Weeks from Deadline

Attend wedding, so no work on electronics. Work on software and sensor integration..

Two Weeks from Deadline

Add $50 (wholesale) power supply, and an over voltage dissipation circuit (10kw, ½ ohm resistor / heating element with high power transistor switching circuit).  Replace FETs from fire and repair all for pre-amp boards.  First wheel is OK, forward only. Second wheel … FIRE (#2).  Fire kills FET board, $1000 motor controller board, and pre-amp boards

Diagnose problem: Hook up scope and find that power supply has some 1 kHz ring that fried FET board, pre-amp boards, and PCI motor controller card..

$1500

One Week from Deadline

Friday – 7 days to go

Order off-the-shelf motor controllers. We want four of them, but only find two, we order them. Order an extra tube of FETS.  Receive a call back from motor controller place at 1 pm Pacific. Turns out they only have one in stock (their inventory computer was wrong.)  Call other companies. The first can’t guarantee shipping. The second won’t sell to us directly. They say we have to go through a local distributor. The local distributor can’t find its <face> with both hands, but wastes our time till 4:30 PM when it is too late to get the parts.

Saturday – 6 days to go

So, we rebuild two of the boards and try the circuit with batteries powering the preamp board.  Seems to work.  Spend three hours trying to find parts to re-build other two boards (fail).  Call a friend who has an electronic manufacturing company, she has the parts and will sell them to us on a Sunday! Fill in the blanks in our autonomous vehicle software.  Run more aggressive tests on the motors … FIRE  (#3).  Kills FETs. Ready to give up, electronics guy has a couple ideas left.

Fire Video (Mpeg 2 megs)

$50

Sunday – 5 days to go

Wake up at 3 am in mad scientist mode.  I come up with an idea to simplify circuit. Wire a forward only motor with a single FET, pulsing plus.  Verify pre-amp wave form, test. FIRE (# 4). The fire kills only a single FET, the least damage so far.

At 9 am rewire, retest, and analyze. Next do full test. FIRE (#5).  Kills FETs, pre-amp boards, and second $1000 motor controller board.  Ready to give up.  Eat lunch, get stubborn, and decide we can get another controller by Tuesday.  We pledge to take Tuesday night to build and test everything and then to take Wednesday off work to video tape everything.

$10

$1100

Monday – 4 days to go

Find the seemingly perfect controllers. Two will do the job at $770 each. (The company even offers a DARPA contestant discount). Call the number, leave a message, and a page, and an email to try make sure we will be able to get them Tuesday. When no one calls back, order them at full price via the web site. Company calls back and then sends email, both confirming we will get them tomorrow.

Tuesday – 3 days to go

The controllers do not arrive. Our electronics guy has an emergency and he can’t take tomorrow off.  Ready to quit, but figure that at this point it would be a shame, because no mater what there can only be two days of hell left.

Wednesday – 2 days to go

The motor controllers arrive at 9 am.  Hook them up, a simple if slow process. Call in a few favors from a friend and get his help for the day.  We do lots of gophering and minor stuff.  Electronics guy stops by at lunch time with food and we finish hooking up the controllers.  Test in the air with notebook computer.  Everything seems great.  Friend arrives with trailer, but trailer is too narrow for vehicle. He goes back to U-Haul.  U-Haul won’t let him tow the larger trailer with his Jeep because a different configuration of his Jeep doesn’t have big enough horse power. His Jeep has the towing rating right inside the door, but they don’t care. Make a really quick RC harness for the car <which I messed up>. Friend returns an hour later will big ol’ trailer behind a U-Haul truck. We have 1.5 hours of light left, and there is traffic now. 

Load vehicle, drive to test field, unload vehicle … disaster (but no fire). One of the wheels slips and now its chain is flapping loose. Screw around for 15 minutes and decide, hey, it should drive with three wheels.  Turn on notebook computer.  Its battery is dead.  Turn on inverter to power notebook. The inverter won’t come on telling us that there is a power fault.  Futz around for another 20 minutes, think we’ve got nothing to lose so let’s see if there is enough voltage for the inverter across two batteries, a surge voltage if you will.  There is a flash of fire (small, so call it #5½) and the inverter is toast.  Try the RC cable, but the cable is no good, and the controllers know it.  Load everything up go home.

That night the gang is all here. We fix everything, charge everything, and test that everything works with the car in the air.  Adjust all the RC settings and run the autonomous programs. We really like those motor controllers. Vow to weld the washers to hold the axles in place.

All three of us have been sick for the last 2 weeks and I’m feeling particularly bad right now.

Thursday – Last day

Friend comes in again, and his wife helps, too.  I am really sick and they end up doing the bulk of the work.  I weld some tensioners in place to prevent any chain from falling off again. They are great, and the right solution.  Plug everything in field-test mode.  The USB hub ($20) connects to USB-to-serial adaptors ($15) and 3-foot serial cables to the motor controllers, which connects to the electric motors.  The setup is the same as last night except that tonight we are running off a power inverter ($80) instead of house power.  Turn everything on. FIRE (#6½). Both serial cables are on fire.  Inventive profanity ensues after killing the power.  We find that we have fried the notebook, the USB to serials, the serial cables, and maybe our beautiful motor controllers.  I have another notebook with 1 serial on it and a 3^rd (different brand) USB-to-serial adaptor.  We do not use the inverter, this notebook is fully charged.  Everything seems to be working in the air.  Get U-Haul truck and trailer again.  Load it up, take 200 pounds of tools.  Drive to test facility (huge lot cleared for development) and unload vehicle.  Run, first doing autonomous tests, forward and backward work great! Turning: it tries, but doesn’t have enough power. How can this be? It should have enough power to drive up a vertical wall. Hook up a diagnostic program (in bright sun on an old notebook LCD screen running on batteries).  And we find that the controllers capping the power at 12% of maximum.  As the motors are rated for more than the controllers, we raise this to 80% of max (because that was easy to set in the configuration tool).  Run more tests, one wheel now working well, but the other wheels are not working well. Also, there is magic smoke coming from one of the controllers.  Hook up RC controller, because what have we got to loose.  One wheel working great, reverse working OK.  Drive a little bit, more magic smoke, kill it.  Shoot information portion of video, load it up, and leave lot at 3pm.  FedEx deadline is 5:30pm.

3:30 PM - 2 hours to go

Hook camcorder to computer, unload vehicle.  We have 1 hour of video taken across the time span of this report, we rewind 20 minutes because there isn’t enough time.  Windows refuses to record from MiniDV giving errors that make no sense. 

4:15 PM – 75 minutes to go

Find out that Windows is attempting to transcode the tape to Mpeg II. Turn that off and start downloading tape.  Keeps stopping, I keep restarting. We end up with four segments. Figure some other settings must be screwy.  Put all segments together, a total of 18 minutes of video.

4:40 PM – 50 minutes to go

Start editing.

4:55 PM - 35 minutes to go

Finish creating five-minute video. Show to friend before burning to CD. Burning to CD fails.

5:05 PM – 25 minutes to go

Copy file to shared directory on network, then copy to a notebook computer. Virus software engages, slowing down the copying process.

5:12 PM – 17 minutes to go

Friend drives, we take the notebook computer, blank CDs and paperwork in car and start burning CD. (“Burn” is just an expression, so no fire). CD works! But we still have to get a form notarized.

5:18 PM – 12 minutes to go

First notary has four sets of people in line looking like they’ve been there for a while.

5:24 PM – 6 minutes to go

Second notary place has two people being helped, but no line. This may work. But no, the next day delivery has already left from there store. But still, we get a rush job on the notarization. (Thank you, mister notary guy).

5:29 PM – 1 minute to go

Head to FedEx

5:32 PM – 2 minutes over

Arrive at FedEx, thinking we can still drive 20 miles through traffic to the downtown location which is open till 7pm, but happily there one truck that hasn't left and we drop our poorly edited, under-whelming demonstration off, and go home to bed.

$2080

Final Notes

We had fuses on every part but only two ever blew. And even those allowed the parts they were protecting to be destroyed.  Also all the fires were electrical, and though we had one person holding a fire extinguisher during every one of the tests it was never used because the auto racing master kill switch stopped all the fires before the person with the fire extinguisher got their hand to the pin. All of the cicuits we used (except the 3am cicuit) were standardized cicuits and tested in the electronics lab on our spare electric motor before we hooked them up to our vehicle.

Another thing to note is this was our second vehicle; we had built a go cart sized prototyping vehicle (Mini SSIK) that we got some a number of tests with, but that vehicle is both too dangerous, and too clearly unsuited to complete in the contest. In testing it had failure cases which included spinning out of control, going full blast is a random direction, and the brakes were not reliable (disc brakes with an electical pull mechanism).

Video of SSIK test run (MPeg 5 megs)

The first, brief segment is computer driven, and the second, longer segment is remote controlled

Total

$4910

Warranties
Hopefully
Will
Cover

$2300

We figure out what our theme music is

I Get Knocked Down
But I get up again
You're never going to keep me down
<Lyrics from Chumbawamamba's song Tubthumbing, our current theme song which I dug up from my music collection>