Robotic VX to compete again!

[fuzzy]

Sadly, it looks like ART's VX is now out of the competition. Wish there were vids/pics of why.

[Techy-D]

Bummer guys. I know you had to have put TONS of energy into this (Blood, Sweat & Tears), but I also know you had a great time and learned a bunch. This is still incredible and even though I won't be able to watch the live updates tomorrow on the other competitors, I'll be looking up info & Pics later.

[Techy-D]

Updates??? I've read a bunch of great stuff about the final race, but how about our beloved VX (It didn't make it to the very end, but it did compete for a while). I'd love an update or two about the VX and what it was able to do or couldn't do, what was the last straw,...

Enquiring minds want to know!

[tom4bren]

I was in a meeting yesterday & the Urban Challenge came up. I asked how the VX did & this was roughly the conversation:

How did the VehiCross do?

Vehi - WHAT???

The Isuzu that UT Austin entered - VehiCross.

Vehi - WHAT???

That thing that looks like it was built for a mission to mars - VehiCross.

Vehi - WHAT???

Oh, never mind.

[fuzzy]

I sent a message to info@austinrobot.com and got this back:

"Thanks for the email! I don't know if my brother Art replied already or not (we both get the emails sent to "info"). I would have updated our website sooner had I known that members of the USAFA were curious about our team! Thanks for asking! The summary is that Carnegie Mellon won 1st place, Stanford won 2nd place and Virginia Tech won 3rd place. Our team was one of the top 21 teams during the semifinals. DARPA had said they would (might) take up to 21 teams to the final event so we were crossing our fingers during the announcement of the finalists. In the end DARPA decided to advance only 11 teams to the final event. DARPA Director Dr. Tony Tether made the announcement of the finalists and he said that he was riding one of the "chase" vehicles during one of the trials when the robot car they were following made a sudden U-Turn and started heading directly towards their car on a full head-on collision course... He said that's when he decided to trim the field from 21 finalists to only 11... Hopefully our team will advance to the final event for the next Grand Challenge! By the way, I can send you a link to our photo album if you are interested. Thanks for the inquiry and thanks for all you guys do for our country!

Jorge"

[tom4bren]

I saw a bumper sticker yesterday that caused me to resurrect this thread. It read:

"In case of DARPA Challenge, this vehicle will be unmanned."

I gotta get me one of those for the VX.

[tom4bren]

Welcome to the twilight zone. Someone emailed this to me moments ago (I used to work with some of the guys on FCS MULE).

Since 2003, the US military has struggled to develop a cargo-carrying robotic "mule" meant to accompany ground troops on foot patrols and into combat. The Pentagon has invested millions of dollars in a range of mule designs featuring a wide range of propulsion technologies - everything from traditional wheels and tracks to articulated legs and snake-like segments capable of "slithering."
Lockheed Martin's XM1217, part of the Army's $200 billion Future Combat Systems suite of technologies, was the most prominent of the abandoned designs. The XM1217 - otherwise known as the Multifunction Utility/Logistics and Equipment vehicle (MULE), based on a custom-designed, six-wheeled chassis, would have hauled up to a ton of cargo while following GPS waypoints. Two of three MULE variants - the ones aimed at fighting mines and carrying cargo - were dropped when the Future Combat Systems program was canceled due to its high cost, leaving an armed version aimed at providing heavy firepower, not carrying heavy burdens.
SEND IN THE MARINES
That leaves the US Marine Corps as the concept's only major champion. To avoid repeating the Army's budgetary and operational mistakes with the XM1217 and other designs, the Marines are emphasizing off-the-shelf automotive components combined with cutting edge - but thoroughly tested - software and sensors. The resulting vehicle might eventually form the basis of a production version, but only after a thorough wringing out. The immediate goal is to "showcase the technology through experimentation," Brent Azzarelli, the Marines' chief roboticist said.
The Marines' Ground Unmanned Support Surrogate (GUSS) mule - jointly designed by the Corps' Warfighting Lab, Virginia-based tech firm TORC Technologies and students at Virginia Tech beginning last year - was given its first big test at the Enhanced Company Operation, Limited Objective Experiment 4 in Hawaii in July. All four GUSS prototypes participated in realistic training events.
"In my opinion, we did real well, we collected a lot of data," and GUSS performed well, Azzarelli says. GUSS is scheduled for improvements to its sensors and autonomy systems and will take part in further Marine Corps experiments over the next couple of years.
The Warfighting Lab, headquartered in Quantico, Virginia, says GUSS "seeks to reduce the dependence of dispersed ground combat elements on external resupply, provide a means to reduce the loads carried by Marines and provide an immediate means for casualty evacuation. The robotic aspects of GUSS enable these functions with no additional manpower and reduced exposure of Marines to lethal enemy actions."
"In short," the release continued, "GUSS lightens the load." But it must do so on some of the world's toughest terrain and in austere conditions, all without costing too much. Otherwise, GUSS could soon join previous mules in the scrap heap of abandoned robot designs.
"We tried not to reinvent the wheel," Azzarelli said. All of GUSS' major components originated from outside the program. The base vehicle is a stock MZ700 6x6 utility vehicle from Polaris. TORC provides all the black boxes from its line of modular automation kits. The algorithms come from Virginia Tech's third-place entry in the 2007 Urban Challenge robot race hosted by the Defense Advanced Research Projects Agency.
TORC and Virginia Tech were a package deal (TORC was founded by university alums). In designing its Urban Challenge racer, Virginia Tech purposefully used TORC black boxes - "so that the technology developed around Urban Challenge could be commercialized quickly," says Alfred Wicks, a Virginia Tech professor working on GUSS.
Wicks' students handled the integration work. "To convert this to an autonomous vehicle, there were a number of modifications," Patrick Currier, a Virginia Tech Ph.D. student, says. "We installed computers, installed a number of sensors to allow the vehicle to operate autonomously and send video back to its operator. We installed a number of cargo-carrying racks, converted it to drive-by-wire, added actuators to the throttle, gear shift and brakes."
ANIMAL BEHAVIOUR
"For the conversion, we did it such that a human operator can get in the vehicle and take over control," Currier added. But most of GUSS' missions are autonomous or remotely operated, project officer Capt. Adorjan Ferenczy explains. "We want to conduct dismounted and mounted ops, point-to-point resupply, casevac, reconnaissance, tele-operations and operation by a live driver."
For interfacing with its human masters, GUSS comes with two options, according to TORC's Andrew Culhane. "The two input devices are a Mac-based [control unit] for mission planning using typical mil-standard symbols for checkpoints, rally points and objectives, and a handheld waypoint device operated as a follow-me device and navigation beacon."
Autonomous operation based on advance mission planning will represent GUSS' basic mode. Compared with previous mule systems, GUSS is more flexible and intelligent in its autonomy. This represents the Marines' effort to address the Army's concern that previous mules weren't "appropriate" for modern battlefields.
Just a few years ago, many ground robots could operate only in environments that were "very structured," Culhane says. They were "highly dependent on GPS and flat surfaces."
"GUSS is designed not be reliant on GPS or roads," Culhane continues. "It uses lidar for determining terrain features and obstacles and, based on the selected mission, if it knows it needs to progress towards the Marines, it will look for the most traversable path to get to the Marines, not relying on high-accuracy GPS, instead utilizing traversability to determine a path from point A to point B."
At the moment, GUSS is equipped only with electro-optical sensors, Azzarelli says. "After LOE4, if we go into development, in fiscal year '11 we will improve perception and add a nighttime infrared camera."
The Warfighting Lab is betting that improved autonomy and the cost savings resulting from off-the-shelf components will justify GUSS' continued development beyond this summer's experiment. "Today, there is not formal program of record requirement for this type of capability," Azzarelli said. "We're hoping to influence requirements generation to provide a capability in the Marine Corps."