The New truck project had originally
been designed and was being built
with one specific purpose and
eventual goal in mind; entering and
completing the DARPA Grand
Challenge so that not only might I
achieve legendary status among
fellow geeks but would also be able
to afford trophy girlfriends when my
boyish good looks eventually fail.
The best-laid plans o' mice an' men
Gang aft a-gley,
An' lea'e us nought but grief an'
pain
For promised joy.
(R. Burns)
behind the bat cave, 4 of the 15 competing teams successfully completed the DARPA challenge. With the
prize money awarded and no plans announced for a future competition, any hope of becoming rich and
famous (or at the very least partially recovering my substantial hardware investment) seem somewhat slim at
the moment.
However, there is no reason for this to also represent and end to my personal goal of building and owning
the mother of all bad assed geek toys from hell.
construction to resume again. Free from the single mindedness of
purpose (and therefore eventual form) that would have originally
been required under DARPA guidelines, I have decided to
consolidate all of my diverse madness into one rolling multi
functional "super project" that besides still attempting to meet it's
primary functional objective (autonomy) will also serve as a
development platform for my other automotive based projects,
transport my telescopes to rural skies on weekends, occasionally
serve as a personal brothel, and be a means of escape should the
apocalypse suddenly befall us.
The original contents of these pages have therefore been altered
somewhat to better reflect current deviations and revisions of the
original plan.
As you may have already guessed, besides natural redheads with freckled chests, I harbour similar unnatural
obsessions with creating robotic oddities and bastardizing delivery vans.
Therefore It's only natural that my latest project involves building a totally bitchin' truck based on a late 80's
Grumman-Olsen P-30 step van platform; the ubiquitous aluminum bodied "Kurbmaster" medium duty delivery
trucks used by, among others, the Canadian postal service. Virtually identical to my previous beast in basic
dimension and Rubik's Cube aerodynamics, but incorporating several fresh volumes of absolutely freakish
improvements and modifications.
Of course It's going to be ugly as sin and twice as tough in the spirit of Mad Max. Nine feet tall, loud, fast, flat
black, and armoured with copious amounts of treadplate. It has to be seriously off-highway capable, able to
suck up secondary and washboard roads at high speed and able to sustain repeated one foot jumps without
damage. Should the situation arise where truck and another object should find themselves at odds with each
other, the former will be required to be the clear victor. Rounding off the Road Warrior wish list is the
capability to ford 30 inches of water.
Grumman produced these trucks with urban delivery duty in mind so the Kurbmaster was never offered with a
four wheel drive option. Substantial modifications will be required to convert the truck and adequately
strengthen the frame and suspension components to ensure their survival under the expected punishing
conditions. A late 80's GMC pickup truck will donate the required front axle, crossmember, and transfer case.
Aftermarket suspension components will be purchased. What I can't buy or steal, I'll fabricate myself. Various
methods of collision protection will also need to be designed and fabricated.
Abbreviated technical summaries: Four wheel drive conversion Body armour
I want moxy: Locomotive stump pulling torque and a top speed in the 75 mile per hour range, but not at the
absolute expense of economy. Gasoline is four dollars a gallon and only going up from here. Although a
moderately built 454 would easily take care of power demands at a very reasonable price, a big block
sucking hydrocarbon shooters at a rate of $0.60 per mile doesn't seem practical. Neither does a $20,000
LS-7 conversion although the thought is sweet. As personally I hold GM diesels in a similar regard to rectal
examinations, the remaining choice may be (perhaps illogically) an inline six.
The old beast could produce a surprisingly capable set of balls from her somewhat stock 292 cubic inches.
I'm considering building up the 292 (4.8L for those that swing that way) with a fuel injection system and ECM
of my own design, with turbocharging being considered. This is a nasty undertaking in itself with fabrication
requiring a lot of work (not to mention math that I unfortunately don't know how to do), but I'm betting against
the odds that if I build a system with a lot of inherent room for adjustment that I can massage it into something
functional through tweaking. Fortunately, there are two things immediately in my favour: By design, all of the
intake and exhaust ports are in a pretty row on one side of the cylinder head opposite the spark plugs, and
secondly I have several acres of unobstructed engine bay on that side of the engine to snake home brewed
and ham fisted tubing with impunity. I really wish I had a clue as to what to expect for output, assuming I can
get it to work at all.
Technical summary: An examination of intake manifold design.
Edit: Developments and recent thoughts on engine choices.
Of course cool tires and rims will be mandatory. Nothing overly insane though; a conservative 33 x 10 mud
and snow light truck radial instead of monster rubber or dubs. I'm also quite partial to plain steel rims and
have been attempting to find out if aluminum moon hubcaps were ever produced for 16" truck rims.
Pure aesthetic rolling evil in the form of treadplate, skid plate, evil bumpers, grille, and quarter protection
aside; lighting, flames, on board generator, frame mounted storage batteries, night vision, LIDAR, RADAR,
video, forward spud guns, microwave pain field generator, pointy shit, and whatever else I can think of that
can be fastened down with bolts and hot glue (except curb feelers and a neon undercarriage) will be added.
Genre styling suggesting that Mad Max met Sweet Tooth and produced an offspring, yet all (strangely)
functional.
Image from Justin's page. I'll send him a
photo of the beast when it's rolling.
microwave oven internals, one could arguably consider that everything to this point is, if not entirely
conventional, at least technically feasible to the average chemically imbalanced high school nerd with a credit
card. Where this project truly differs, is that I intend to build a completely autonomous vehicle that will be
capable of navigating and following the roadway without driver input while avoiding obstacles, ditches, Bugs
Bunny style cliffs, and at least twenty percent of slow moving pedestrian mammalia. Simply because I think
that I can.
Gentlemen, I shit you not.
Several years ago my employer and friend Stephen Wilson related an article about a robotics competition
where driverless vehicles competed against each other in a race across the desert for a 2 million dollar prize.
Further investigation revealed the competition to be the "Grand Challenge" sponsored by the Defence
Advanced Research Projects Agency. A development organization for the Department of Defence. The
annual competition is open to anyone who qualifies and offers 2 million dollars to the first entry to complete a
175 mile course across the desert in under 10 hours. The vehicle must be totally autonomous and navigate,
avoid collision, and map terrain by itself without any external influence or signals that are not publicly
available.
In 2004, over 40 vehicles offered by teams comprising of people ranging from screwball inventors to robotics
academia competed at a qualifying event where DARPA officials attempted to separate the insanity from the
potential. 15 teams with diverse platforms ranging from a modified ATV to Team Caltech's very sexy ALICE
Ford E-350 van conversion were chosen to compete in the Grand Challenge. When the Mojave dust had
settled, the prize money remained unclaimed.
Grand Challenge event photos - 2004.
My original published intention in September 2003 was to attempt to have my creation entered into the 2006
qualifying event. I wasn't even close.
Even to date (April 2006) the majority of my plans still exist either untested on paper, in my imagination, or
scattered about my laboratory in pieces. A suitable vehicle wasn't even acquired until April 2005. With the
belief that the prize would remain unclaimed for the foreseeable future I set my goals toward completing my
entry tentatively for 2008 or 2009. However, on October 9, 2005. The prize money was claimed by Team
Stanford who created the first of the four driverless vehicles that completed the 131 mile course through the
desert well within the required 10 hour time limit. I find it truly ironic that as the feasibility of autonomous
driving was being realized some 3000 miles to the south, A woman blasting a red light at a Mississauga
intersection totally destroyed my car.
Discussion: Was this goal ever truly achievable?
FAQ:
The future may very well include the possibility of another perhaps more difficult challenge. Although the
DARPA mandate has been satisfied and there are currently no plans for further events, the EE Times has
quoted Intel's Gary Bradski as somewhat prophetically (although intended as levity) stating "Now we just have
to teach them to drive in traffic".
Although I have no plans to attempt a feat such as autonomous rush hour city traffic at the moment, nor do I
believe it's remotely possible with current levels of computer technology, work will continue on my own
driverless creation.
The answer unfortunately is seemingly nowhere.
I've come to realize that for every week of design or theoretical work completed, I have been adding months
of eventual construction. It is a trivial matter to build and program a 30 pound wheeled robot to navigate a
warehouse full of obstacles at human walking speed, but the dynamics involved do not lend themselves to
simply being scaled to vehicular mass and speed. Interestingly enough, they would given a flat closed course
and perhaps an Abrams tank.
The truck is being built. The mechanical and aesthetic repairs and modifications of the body and chassis,
although far from complete, are well underway. Many electro-mechanical components and assemblies, such
as various actuators for accelerating, braking, and gear selection, steering drive motor, and feedback
sensors are being installed without a completed overall system design as the straightforwardness of their
individual function will not require changes as revisions are made. Likewise the air system, power generation,
storage, regulation, and chassis management computers can be installed and tested as stand alone
components.
Plans call for the chassis to be ultimately controlled by a pair of relatively "dumb" computers, one of whose
job will be to simply convert chassis operational output data (such as the position of the steering wheel,
speed, and cornering forces) into 24 bit "words" that will be interfaced to "smarter" computers doing the
thinking. The other will convert data from similar words into chassis control input commands (such as steer
120 increments left and reduce throttle by 50 increments). This approach to "opcode" oriented system and
multi program networking will hopefully allow me to build the entire system in functional blocks, that should if
nothing else, allow me to feel as if I'm always progressing forward.
Eventually, hundreds of sensors will be installed on "cages"; large aluminum frames that can be easily
unplugged and removed from the vehicle. The idea of leaving a pair of LIDAR units worth more than my
house in a tavern parking lot is frightening.
The final, and probably the longest, development stage will be to actually make it all work together. As of my
last revision in design I will require a total of 13 computers and I wouldn't bet against this number increasing.
Besides personal taste, The Kurmaster platform was chosen because there is sufficient space for a
comfortable work area on board for developing and debugging the software that will be required to convert
huge amounts of sensor data into useful driving commands. My intention is to initially complete a (manually)
drivable vehicle that can be taken to remote areas on weekends so that I can program in peaceful
surroundings with my dawg, and not run over things that scream while testing the code.
I have recently decided that any attempt at estimating a completion time is hopeless. I intend to have the
truck drivable by the summer's close (2006) and would be pleased to see an initial "parking lot" driverless run
in under two years, however the latter will not only hinge considerably upon the former, but also require a
lifestyle change. I work very long hours, and live in a constant state of paranoia worrying about further
munchkin attacks damaging the truck while I'm at work. If I can get the beast rolling, legal, and in a form that
doesn't attract too much attention (for now) this summer, then it can be transported to work on a weekly
basis, not only alleviating my stress, but freeing an additional 25-30 hours of time during the week to develop
her. Also, once the major interior fitments are complete and the cabin is better sealed and insulated from the
environment, the long winter months can also be put to use in this regard.
In the mean time, the project will likely continuously grow as my sanity wanes. I will almost certainly manage to
sidetrack myself by stuffing some other "device under development" into the "final" design. It just happens.
The absolute most wonderful thing about being me, is that It's never a question as to whether or not it should
be done (or whether it makes any sense), but simply a matter of can it be done? My "highway paintjet" (see
my projects page), originally nothing more than a thought experiment a few years ago simply has to be
incorporated onto the bat truck because it can be, and as illegal as it may be, watching my friend Pat Taylor
do mega carnage with an Idaho spud propelled by compressed air inspires me to sling a fender whumper of
my own under the beast for those bitchy traffic days. So many gadgets, so little time. Flamethrower anyone?
Finally, on a more personal note. I would like to thank all of the members of the Dutchforce Electronics Forum
that will undoubtedly assist me in this undertaking when I get stuck on some detail yet to be discovered. May I
make all of my bother Nerds, Geeks, and Dorks proud.
Construction details, diary, photographs, technical writings, drawings, and other aspects of the
entire project can be found at the link below.
J Tyson. 2006.
