Jalopnik
[Jalopnik's desert stringer Curtis Walker is following the DARPA Grand Challenge, an annual race of autonomous ground vehicles, with a serious case of the techies. We'll have his reports today on the National Qualifying Event semifinals, like this one, as soon as he can type them into his digital audio-visual receipt and transmission unit. - ed.] Among the numerous newcomers to this year's DARPA competition, Massachusetts Institute of Technology seems poised to win the imaginary award for most gizmotastically outfitted. Armed with a cool million in seed money for development, team MIT went all out with the sensors and CPUs. All told, they've got 11 Sick Lidars (Light Detection and Ranging) units, five optical cameras, 15 Delphi long-range radars, GPS and a 40-core supercomputer to run them. All this power comes at a price in the form of electricity required and heat generated. To address this, they installed a 6kW generator and a 2kW roof mounted AC unit. Perhaps even more impressive is that the 100,000 or so lines of code running on the system were all written for this event. [Next: Team Lux, and gallery.]
At the flip side of MIT's absurdly outfitted LandRover is team Lux's Volkswagen Passat, a joint venture between German supplier of automotive sensors Ibeo and parent company Sick. Hands-down winner of the nonexistent prize for stealthiest install, their diesel grocery-getter has a mere three sensors; one on each corner up front, mounted behind black Plexiglas in the bumper and a prototype model mounted in the center of the read bumper. Data from these units is collected and processed by four computers in the trunk. Even the actuators for steering and velocity are stealthily mounted. Apart from the big honkin' kill switches on the read windows, one might miss the fact that it's a robot car at all.
It's hard to pick favorites in an event like this, and looking at this brief comparison shows why. Despite their differences, both vehicles did well in Saturday's qualifying runs. [DARPA Grand Challenge]
Comments
I love these sweet competitions. The x-prize and DARPA challenge are pushing technology forward in an extremely interesting, competitive and cost efficient way. If only American Idol could produce such good results...
If Team Lux can do rouhgly as well as Team MIT with less than 1/10th of the complexity, then their approach is definitely preferable. It's not about geekoid bling, it's about reliability.
Team Lux should've tried to pick up a cheap used Phaeton. I'm sure the Passat has more room for boxes of unidentifiable computer bits, but the self-driving bit would've suited the Phaeton so well.
You couldn't find two more extreme examples. I'm with the Lux--what's the point of a Robocar if it screams "Robocar" -- sitting duck target for a shoulder rocket?
They've got to work first. And not that I'm saying the Lux car won't work, but you can only navigate as well as the sensory information you have. So the general rule of thumb is gonna be more sensors is better, but we'll have to see how it pans out. I agree that if they work equally well, then the more elegant solution is the better one, but the first step is to find a solution that even works, then you can go to work on simplifying/shrinking it.
I have a question, what do they do wit the cars after the race is over?
You would think that a group of MIT engineers would have the sense to not pin all their ambitions on a Land Rover being able to operate reliably. What good is a million dollars in sensors if the coolant goes shooting out of the engine five miles into the course?
@ dculberson -
agreed. They should have gone with a Superior Consort instead. Those things are bulletproof.
Sick Lidar unit, yo! <-couldn't be helped
Team Lux's Passat wins, if only for not looking like a gamer's overclocked rig..
Just needs a little more Evil.
@Novaload: Ummm, misunderstanding of the design goal. This technology, in a several-generations-further version of course, will be the basis for a new family of supply delivery trucks. Not the high end of military action, but the low end...the behind-the-scenes delivery convoys through potentially hot urban areas where it really makes no sense to expose human drivers to IEDs. All such trucks intentionally will be readily recognizable for what they are. It's militarily desirable for the bad guys to know the trucks are robotic. It may not make sense for the bad guys to spend their $$ on ordnance to kill a truck...they know that in the end, they can't make us run out of hardware. So far, it's been easy for them to score psychological points on us by taking out a human truck crew. The design intent is to eliminate that way of scoring on us.
@JWilly48519:
In regards to military desirability of destroying robotic trucks - I'm not so sure you're totally right. The first thing that came to my mind was that robotic trucks in a combat zone, while not becoming the targets of IED's in the traditional sense, would have the potential of being set up as traps.
A hostile force could potentially trap a robotic truck with barriers, then rig IT with IED's for the people who come to un-stuck the truck, or even attach an explosive device onto a robotic truck while its moving, then detonate it when it gets to its destination.
I think that if the design intent of robotic trucks *really* is to take human drivers out of harms way, then the designers/advocates/military better be thinking about the ways that robotic trucks could be used against them.
Start a discussion:
Login with your username and password below. Or comment on this post via email.
Forgot your username or password? New User?