CHAPTER TEN

The Challenge

THE DARPA GRAND CHALLENGE CHANGED RED Whittaker’s life and, to a certain extent, turned the Field Robotics Center upside down. To say that Red Whittaker was elated and ready to jump in and declare himself on behalf of the university, the city, the entire robotics world, is an understatement. Folks at the Robotics Institute instantly recognized Whittaker’s sudden transition—his fiery eyes, his manic dashing about the campus, visiting colleagues and talking up the impending event. Whittaker had been transformed back to his old self by the lure of the DARPA Challenge. Groundhog had been a warm-up, getting him back into the heat of the game. Now the old soldier was blessed and electrified with a new mountain to climb—an impossible idea to contemplate and to achieve.

When the complete details of the competition were announced, Whittaker was present in Los Angeles near the scene of the race’s proposed starting point, to hear it firsthand and to promise to lead the first team to jump in. Returning to Pittsburgh, pumped with exhilaration, however, he was appalled to learn that Carnegie Mellon and Robotics Institute administrators had decided to opt out of the Grand Challenge—a practical decision to the folks who controlled the purse strings.

As part of the rules of the competition, DARPA prohibited the use of equipment or technology that had been supported by government money. At the time, Carnegie Mellon administrators felt that an unmanned ground vehicle (UGV) being developed at the National Robotics Engineering Center would possess similar capabilities as the autonomous racing machine DARPA was seeking to have created through competition. Duplicating the NREC’s efforts seemed pointless—certainly not the kind of project that would keep a small private university with a moderate endowment like Carnegie Mellon, sustained by research money, primarily from governmental sources, in business. And the $1 million prize would not nearly compensate for the resources required—a minimum of $2 to $3 million to be competitive.

DARPA had purposely included this nongovernment-money stipulation to entice talented people who don’t usually deal with the military industrial complex, as do Carnegie Mellon, MIT, and the Jet Propulsion Laboratory. “We are inviting little mom-and-pop folks out there to help spur advancement and take us where we need to be,” explained Jose Negron, the DARPA race project manager. By “where we need to be,” Negron is referring to the congressional mandate stipulating that one-third of all ground combat vehicles operate unassisted by 2015.

In its search for new ideas and approaches from “fresh blood,” as Whittaker might have put it, DARPA was also making an attempt to even out the playing field. All of the racing robots would have to navigate and accelerate autonomously. The humans behind the robots would receive global positioning system (GPS) coordinates two hours before the starting gun sounded on March 13, 2004, race day. Each team would enter the data in their robots’ computers and then turn their robot over to DARPA. The robots would be, theoretically, autonomous at that point.

When the Challenge was announced, most roboticists, including those at Carnegie Mellon, reacted by dispelling any notion that a robot could win, or even finish, the race. A half-dozen major problems were cited. For one thing, robots function poorly in “negative terrain,” like ditches, ravines, and creek beds. Vision systems are much more reliable on flat ground with unaffected or uncontrolled lighting. The race in the desert would churn up a lot of dust, and the robots would be forced to make important decisions about the source of that dust. Was it from a windstorm, from their wheels, or from a challenger coming up from behind? How to react, in each case? If from a challenger, the problem is more serious. Thus, the navigation program must be written to avoid obstacles on the course (rocks, trees, barrels, etc.) while being aware of the proximity of other robots—and then to make split-second decisions to pass or avoid them.

The algorithm that allowed Groundhog to stop or think for five or ten seconds before deciding on direction in the Mathies Mine would be of no use in a race where a vehicle averaging thirty miles per hour must make decisions instantaneously. To win the $1 million, the robot would have to complete the course within twelve hours and pass through specific narrow corridors or checkpoints from start to finish.

If Groundhog motivated Red Whittaker to wade back into competitive waters, then the DARPA Challenge made him dive deep, despite being confronted by ambivalence and resistance from most of his colleagues. It would be wrong to say that he didn’t care that his home institution was reluctant to support him, but the resistance increased his determination and desire. His own first challenge was to get support for the Challenge, and he hit the ground running by reaching out to his former students now working for major corporations and to his contacts in the media to announce his intention. Polite and unusually politically astute, he was careful not to directly involve or exclude Carnegie Mellon. A group he was calling “The Red Team,” he said, was taking up the DARPA Challenge gauntlet independently, although he would use Carnegie Mellon as the Red Team’s home base and some of the facilities at his disposal.

But even though he knew in his heart that he would not give in to negative pressure and walk away—that this was the moment of his destiny, the moment he had been waiting for to get back into the game, full steam—he wrestled in his characteristically elaborate manner with whether he was doing the right thing.

“I believe that this will shatter the threshold of what the world views possible,” he told me. “The prevailing view is that this is an unwinnable race, but does the fact that a mountain is unclimbable deter people from trying to climb it?” And then, in the very same conversation, a few minutes later, he would doubt his intentions, asking himself if the idea of committing energy and resources to this impossible competition was “self-indulgent.” Was he being caught up in the self-serving excitement of the moment? he asked himself. And then, like a pendulum, he would swing back again. “But vision without implementation is irresponsibility.” During those early days he went back and forth like this with many others. A reporter, Bruce Steele, who had written an in-depth feature article about Whittaker for Pittsburgh Magazine, commented later: “He seemed to enjoy viewing himself at a distance and pondering his own motivations, as if even he found himself to be a fascinating character.”

The buzz generated by the local publicity, combined with Whittaker’s persistence, eventually pressured Carnegie Mellon into pledging additional facilities, public relations, and moral support. Though Whittaker would still have to raise an estimated $2 million in cash and resources on his own, he was not deterred.

In addition to the old friends and protégés he had already checked-in with, he began cold-calling engineers, software scientists, corporate executives, and venture capitalists, enticing them with his enthusiasm and promising a previously unimagined adventure—technological swashbuckling” to be sure. It took some time, but Whittaker gradually lured in prestigious research organizations, including Intel, Boeing, General Electric, Alcoa, and Google, to contribute money and personnel.

Getting an idea in his head and pursuing it, despite ambivalence and opposition, was predictable Whittaker behavior, as were the methods he used to transform his project from dream to reality or, as he said, “vision to implementation.” The entire scenario was a near-repeat of his Groundhog saga, but on a much larger level, beginning with his quest for “fresh blood.” Initially, graduate students were pretty much out of his reach, since they required funding. But there was another resource, the rawest of the rookies, what he termed “The Children’s Crusade,” a truly accurate description. As with his mine-mapping efforts, Whittaker introduced a robotics course, this time aimed at undergraduate students:

 

New Course, Summer 2003

FUNDAMENTALS OF MOBILE ROBOT DEVELOPMENT

Applications like space exploration, border security and desert racing motivate mobile robots to operate in barren terrain. Challenges include boulders, streambeds, slopes and cliff sides, and these become more difficult to negotiate with increasing speed. The issues and systems that pertain to desert robots are unique, and the technical state-of-art is insufficient for the functionality and rigor needed for future systems.

The summer ’03 offering of Fundamentals of Mobile Robot Development will convey the issues, technologies and complexities of Robotic Desert Traverse through lectures, exercises and field experiments in a learning-through-doing environment. The course will convey the distinctions of robotic racing technology including topics of specialized sensors, mobility, mapping, game theory, high performance computing, and self-reliant systems. The course will formulate and develop a complete desert race robot and demonstrate basic system performance.

Course context is the CMU Red Team bid for victory in the robotic LA-Vegas Race for a million dollars in March ’04. CMU is committed to the top.

As with the Groundhog project, Whittaker also initiated an e-mail Race Log sprinkled with up-to-date information about the project along with a heavy dose of inspirational messages to buoy his rookie troops:

The strength of the pack is in the wolf and the strength of the wolf is in the pack

—RUDYARD KIPLING

Let’s hunt, kill and eat together

—RED

And he established a timeline of days, beginning near the one-year mark and counting down:

A year from now, we will be racing. Many believe that the Challenge defies possibility, and that this race is un-winnable. Whatever the fate of our robots, our spirits will triumph.

And he convened pep-rally-like meetings—bringing his growing retinue of students together with representatives from an increasingly impressive list of sponsors from the private sector. At the first official Red Team meeting, he stood up and proclaimed the moment as “the Golden Hour.” “We are already a team. It is the first time we have seen each other, but some of us know that even now we are already a team.” And he walked around the room handing out nametags for everyone to fill out. “The highest level of commitment is to use each other’s name as we put a team together.”

Near the end of the meeting, he went around the table and confronted them, one by one, with direct and telling questions. First to the students from whom he was seeking commitment:

“Are you in?” he barked.

“I think so.”

“If you think so, then you have to leave the room. Are you in or out?”

“I’m in.”

“OK, we are now hunting as a pack. Shame on us, if we can’t kill anything.”

And he was, in fact, “shameless” (his own word) in asking the corporate leaders around the table for money.

“Lindbergh crossed the Atlantic and won the Ortiz Prize. He cashed the check ($50,000), but he didn’t do it for the money. Who has R and D money for this project?”

“We have budgeted R and D for $250,000,” an executive admitted.

“Would you be good for six figures?” Whittaker replied.

The Red Team was the first to enter the DARPA race and Whittaker’s prominence elevated the cachet of the competition from the very beginning. Other prestigious robotics groups like Stanford and MIT felt, as did the Carnegie Mellon higher-ups, that there was more to lose than gain by competing. There may have been dissenting voices in those institutions, but none so obsessed as Whittaker’s in countering the dissent. Dollar by dollar money was raised, more than a million in equipment, services, and cash that first year. Whittaker, a legend in his own little world of robotics, suddenly became a national media figure.

As the race grew closer, Whittaker and the Red Team were featured on National Public Radio, in Fortune magazine, Scientific American, Popular Science magazine, Wired, and The New York Times. Both the Discovery Channel and the History Channel were on hand on rollout day, when Sandstorm, a stripped-down and reengineered Humvee, was unveiled on campus to the public and the media three months before the race.

The event was elaborate and festive. Optimism was peaking. Whittaker had received the national attention he craved and the money and support he needed—far beyond his initial expectations. He had, in fact, achieved what everyone had really thought to be impossible. He was favored to win the race. A legion of disciples, perhaps a little more well-groomed for the occasion than usual, with flashy Red Team racing shirts, were on hand, exhilarated, joking with great bravado, about their leader’s obsession and their lack of sleep.

Phil Koon, an engineer on loan from Boeing, revealed that he had slept in the apartment he and his wife had rented three months ago for his stay in Pittsburgh “for only two nights—total.” Koon brought a sleeping bag to campus, while Chris Urmson, who put his Carnegie Mellon dissertation in limbo for Sandstorm, reported that he always thought he needed a lot more than four hours a night to sleep, “but Red showed me I needed a lot less.”

The Sandstorm rollout event in December 2003 in their headquarters complex, a building donated by Carnegie Mellon, was classic Whittaker in the way he worked his audience. Towering over everyone in his conservative blue blazer and khaki trousers, Whittaker dominated, pointing, joking, injecting information and insight in an engaging, charming never-ending, medicine man–like banter. His gaze swooped through the room, from face to face, like a spotlight, connecting with everyone for an electric instant. Periodically, he stopped to ask a question or make an observation about their Red Team work, always prefacing these questions with an infusion of compliments so that every respondent was positioned in a positive light. He had chosen a Humvee because of its durability and because it had been recommended by racing experts he had consulted.

For the people who saw him for the first time, he was mesmerizing; for those who knew him, he was simply engaged in a “Red monologue.” But for everyone, new or familiar, he was relentless in his quest to charm, motivate, and unabashedly manipulate people to do more than they ever thought possible. The dramatic showmanship hit its peak when a tarpaulin was lifted hydraulically and there, suddenly, like magic, was Sandstorm, a glittering red, converted 1986 Humvee with a domed top and banks of sophisticated computers.

As onlookers examined Sandstorm, Yu Kato, a tall, longhaired graduate student from Japan, who had written some of its software, confessed that Sandstorm had become his passion. After starting as a volunteer in September, he brought his rice cooker to campus so that he didn’t have to go home to eat. Struggling with his English, Kato, in loose black pants and open-toed sandals, sighed and smiled: “I want to sleep, but Red won’t allow. He want me to work, fast, like I am running a hundred-meter sprint, and then, when I stop, exhausted, he want me to sprint a hundred meters again.”

“Young people can work all night,” David Wettergreen told me. “And they have less perspective on when they should stop. They overcome lack of knowledge and experience by just putting in long hours. It helps to not have a family at home waiting for you.” Wettergreen discounts the notion that young people work to please their mentors. “I think that the challenge of creating something unique or the fulfillment of seeing something you made work are stronger and more sustaining motivations. If it is a matter of pleasing anyone it would be their teammates and peers”—an observation that may contradict Yu Kato’s motivations.

“And do you do it?” I asked Yu Kato. “Do you run another hundred meters?”

“Well,” said Kato. “For Red, I try.”