At the headquarters of Fantastic Comics in Manhattan, Iron Man comics editor-in-chief Tom Breft briefly introduced me to Tony Stark: MIT graduate, star scientist/engineer, wealthy heir, playboy, and alcoholic. Later, bad guys kidnap Stark and coerce him into building a deadly weapon for them. Instead, he ended up building a metal armor. After escaping, he comes to his senses, improves the armor, and transforms himself from a selfish genius into a true superhero.
Iron Man outruns jet planes, lifts 1,000 tons and breaks into heavily fortified computers. Like a pipe dream? Of course, Breft says, that's the beauty of fantasy, "Iron Man must always stay ahead of reality, or he'd be a fossil by comic book standards."
In 1963, when the character first appeared, the Army was also envisioning its own Iron Man. That same year, U.S. Army weapons researcher Sergei Zarudny published a report describing his design for a wearable machine exosuit that would give the wearer Hulk-like strength, but the technology to make the idea a reality didn't yet exist. With the exception of a few non-military designs, the prospects for a truly super-powered exosuit were slim until 2000, when Darpa began a seven-year $75 million research program on mechanical exoskeletons. By then, a handful of mechanical exoskeleton proponents believed--including U.S. Army Colonel Jack Obusek--that the technology was finally catching up to the conception. Since 1995, Obusek has helped advance exoskeleton research. As sensors increasingly become smaller and more versatile and processors speed up, he said, he and other supporters have reason to believe that mechanical exoskeletons have the potential to become a reality.
But Darpa's ambitious goals are impractical to anyone who looks at them. It wants a miraculous machine that will allow soldiers to run for days on end lugging hundreds of pounds without tiring; that will give them the dexterity to operate weapons that normally require two men to navigate; and that will allow them to easily evacuate a battlefield with two wounded comrades on their backs. They demanded that the mechanical coat come with armor that would make it invulnerable to enemy fire. They even wanted it to help soldiers jump higher. In short, they wanted the Iron Man of the comics.
Before launching the program, some of the consultants Darpa consulted immediately pointed out the impracticality of their vision. Says Yvelacine Garcia, the University of Cornwall engineer who was initially in charge of Darpa's Ironman program, "Half of the people I asked about it were convinced by it, and the other half thought it was simply a waste of time, money, and resources." Those who poured cold water on it weren't wrong, he added, "It really is a daunting challenge." The mechanical exoskeleton would require a lightweight power system capable of supplying electricity for days on end; it would also require small, powerful artificial muscles; a complex system of motion control. It must also be agile and responsive.
The exoskeleton must become the soldier's mechanical shadow, able to read his every move and mimic his every action in such a timely fashion that even milliseconds of hesitation can be burdensome, making the soldier feel as laborious as walking on water. The mechanical exosuit's sensors must be able to read every slight movement applied to its entire body at a rate of thousands of times per second, and its microprocessors must be powerful enough to convert that data into commands transmitted to the mechanical limbs in time to harmonize them with the actions of the wearer inside.
To solve these problems, to figure out how to fit this system into a machine that combines speed, agility, strength, and endurance, would take a genius of the Tony Stark variety. But that person is not a weapons designer, but someone who has made a career out of building mechanical dinosaurs. Steve Jacobson has a colorful resume. Over the past 35 years, his credits include an 80-ton robotic dinosaur, the fountain at the Bellagio Casino. Yet, in person, he looks more like an affable professor than a science nerd. He is tall, broad-shouldered, straight-backed and gray-haired. Before introducing me to XOS, he takes me on a tour of what he calls the "Tunnel of Terror". From the outside, it looks like a dentist's office, but this cavernous room is actually the headquarters of Sakos and the R&D department of the University of Utah's College of Engineering. He often teaches here. Despite having built robots for a number of notoriously picky customers - he implies that the Disney company is as demanding as the U.S. military - at heart Jacobson is still an academic researcher . He calls his mind a friend to play with, and he seems more concerned with problem solving than with the eventual application of the solution. After passing a humanoid robot that plays ping-pong, he stops in front of a pair of singing mechanical hornbills. They were built for a local hotel. The difficulty was making them move as lifelike as real birds. He says, "We only take jobs we want to do because they pique our interest."
He speaks eloquently, and the conversation moves from engineering to energy-efficient biological systems (carrot-powered people!) His talkativeness can be misleading. In fact, he's a man who prefers to keep secrets, rarely giving interviews to the press, and not even willing to divulge his age. During the tour, he pointed to a small unmanned ground vehicle (a new design for exoskeleton legs) and explained it excitedly. At the end he instructed me not to mention it in the article. I think his concern may be because so many of the projects are funded by the military, and on the other hand, he's like a magician who doesn't want to give away too many secrets.
Sarkos's research projects - which also include prosthetics and nanomotors - seem varied and unorthodox. But it's the breadth of interests that makes Jacobson the perfect candidate to take on the exoskeleton challenge, says Ephraim Garcia. He has proven himself in both software and mechanical engineering, and, more unusually, he is constantly inventing new things as needed. "He can design his own drives; he can design control systems; he can design a machine and all its parts." Garcia said. Such an all-around genius was definitely what was needed to design XOS.
"When designing something like an exoskeleton," Jacobson said, "there are 25 subsystems, and you can't move on to the next step until you complete them. While the two main design goals are strength and endurance, it also has to be able to do 75 different things." Of all the robots he has designed, XOS is clearly his favorite due to the greatest challenges and problems, and he considers it his "son." "None of the others were so ambitious. Nothing else requires such a complete and independent system, such strength, speed, endurance and sensitivity. In 2000, Sakos applied for a Darpa investment. Jacobson thought he had found the answer to Darpa's bounty problem. That problem was how to get operators to interact with robots. To confirm his hunch, Jacobson asked Jon Price, the company's photographer, to help him run an experiment with his daughter.
The experiment had Price play the exoskeleton and his daughter play the internal operator. She turned her back to her father and stood on his feet, her toes pressed against his. They held hands to help balance. She begins to walk toward the front. Price's task was to stay in sync with his daughter, keeping his feet under her at all times. After a few minutes, they were moving like one person. His daughter was in complete control of decision-making -- how fast to go, when to turn -- and Price was just trying to imitate her, one step at a time.
The demonstration proved to Jacobson that it only takes a few points of contact -- like feet and hands -- for an intelligent machine to understand the operator strapped to it's movement intentions and act in tandem. While the theory is simple, the practice is quite difficult. In the course of completing XOS, Jacobson and his team redesigned the microdrive, improved the pressure sensors, invented more efficient hydraulic valves, and even designed the robot's aluminum footplate. But it was the control system, which the designer called "Clear the Way," that held all the pieces together and was the centerpiece that elevated yet another one of his robots into a "superhuman suit. Obusek, who has tried on XOS himself, agrees: "A little bit of weight and the human body gets tired easily." But the XOS control system reduces the burden on the body to near zero.
It was this control system that enabled demo operator Jameson to lift a 200-pound barbell 50 times in a row without his heart rate accelerating. When he pulled the barbell off the rack, sensors in his hands immediately detected the change in torque. Had he not had the help of an exoskeleton, the sensors would have shown that he was trying to pull down 100 pounds of weight with each hand. But, Jacobson explained, the goal of the system is to get the pressure felt by those sensors close to zero so that XOS can function.
The sensors in those hands send their measurements to a central processor hundreds or even thousands of times per second. This system feeds the data into a series of formulas that calculate the orientation and movement of the exoskeleton's arms, legs and back. Eventually recognizing that Jameson wanted to put his hands down, it calculated how each artificial muscle within each joint would need to move in order to mimic his movements. Jameson never felt the least bit burdened because before he actually exerted any force, the system had already commanded the robotic arm to take the barbell on his behalf. After the lifting exercise, he unloaded the XOS without any sign of breathlessness. I asked him how he felt. "Great," he said and shrugged.
The XOS that Jameson demonstrates is about version 4. Jacobson takes me on a tour of a room in which the first 3 prototype mechanical exosuits hang like puppets on shelves. I was immediately reminded of Iron Man's "Armor Room" - the room where Stark kept his suit. The first suit was built in 2002 and was unpowered. Sarkos' team built it to prove that an exoskeleton could move as well as a human body. Jacobson strapped an engineer inside the suit and had him attempt various complex maneuvers, such as kicking a soccer ball, running, and climbing into the cab of a car. Through this series of experiments, they confirmed that the right joints were used in the right places.
Making those joints open and close with the right speed and force was more difficult, and in 2003, Sarkos began using water pressure to drive actuators to make artificial muscles. This approach was not his first. In fact, another exoskeleton researcher believes the XOS's reliance on water-pressurized devices will eventually lead to its failure. The engineer, who wished to remain anonymous, has not seen XOS in person, only a news clip on YouTube. He said the water-pressure unit is too power-hungry. Electric actuators are better, he argues, because they consume energy in line with the action. But Jacobson impatiently rebuts this criticism. "Do you like your automobile brakes? Do you want the landing system of the airplane you're flying in to work properly? They're both hydraulically pressurized." He later added that he had solved the energy-wasting problem. But he wouldn't explain the details, except to say that Sarkos redesigned the valves that control the flow of fluids so that they are activated only when they are needed, so energy is consumed only when the mechanical suit is in motion.
While impressive in the weightlifting program, the XOS doesn't meet all of Darpa's goals. It can't make you dunk in the air, it can't help you run faster, and it can't turn you into Hercules (Hercules). But, Obusek said, one of Darpa's goals in the first place was to see if any of the items on its wish list could be realized. Of the three teams that participated in the program (Sakos, Oakridge National Laboratory and UC Berkeley), XOS emerged in 2005 as the closest to what the Pentagon envisioned as a full-body exoskeleton approved by the U.S. Army to move into the next phase of research. Sakos has been awarded $10 million in Army funding to cover two years of research.
Jameson donned the XOS again to demonstrate a relaxation workout. As I watched the 150-pound machine exosuit mimic his every movement like a shadow, while only six of their joints touched each other, and imagined the vast amounts of data flowing between each sensor and the central processor in seconds, the scene in front of me was as exciting as the special effects in the "Iron Man" films. I could hardly resist envisioning Jameson flying through the roof. But that's not going to happen. To break out of the cave, you first have to cut the wires to the XOS. The world of exoskeleton researchers is small, secretive, and characterized by a lot of back-and-forth. Although, there may be no understanding of how a rival device works, there is still no shortage of cynicism. The most common attack is, "You asked him how he plans to power it." XOS and the other two more leading exoskeletons in the U.S. labs attempted to attack this conundrum from different angles. Jacobson decided to build a fairly capable suit first, and then drill down on how to power it for 4 to 24 hours (the minimum power requirement suggested by Darpa). In all the demos I've seen, Jameson and XOS are connected to a hydro pump, through which they get power from an external source. The suit can be battery-powered, but it only works for 40 minutes at a time. Two other exoskeleton researchers - Prof. Hugh Hear of MIT and Hamayon Kazrouni of UC Berkeley - have begun to address the power issue .
Hurl is trying to build a foot-powered machine that saves as much energy as possible -- it requires only 2 watts to start up, the equivalent of the power consumption of a portable radio -- yet support 80 percent of the 80 pounds of weight the wearer carries. Designed to affect the wearer's gait, it uses slightly more energy when worn than when not. However, Herr believes that in the near future he will be able to improve the mechanics so that the machine will ultimately help save rather than drain the wearer's energy. Ultimately, he envisions the device being used for recreational purposes, with people wearing it on weekends to go hiking and running all day without feeling tired. If Herr's vision is still relatively distant, Kazrouni suggests to me that he's already halfway there when it comes to solving the power problem.
Kazrouni says his Human Load Carrier (HULC) lower-limb exoskeleton can work for 20 hours straight. It allows the wearer to carry 100 pounds on his back, but consumes 15 percent less oxygen.
Kazrouni's device cannot yet be demonstrated to the public. He would only reveal that the system works like a hybrid car. Hybrid cars convert the energy generated by braking to recharge their batteries, and HULC utilizes the energy sent back from the ground when the walker switches feet. The very act of walking is enough to generate a constant stream of kinetic energy. He has been awarded a $2 million grant from the National Institute of Standards and Technology to improve the system. Eventually, the HULC will help people with mobility impairments resume walking. "This is not a war machine," he said, "our machine may replace the wheelchair."
The XOS's strongest competitor is also a medical device, but located across the Pacific in Japan.In 2004 Japanese roboticist Yoshiyuki Yamagai founded a company called Cyberdyne (the same name as the company that led to the robotics revolution in the movie "Terminator") to market his full-body robotic exoskeleton, the HAL-5.Instead of using XOS-style pressure sensors. Instead, it attaches sensors to the wearer and receives signals from his or her muscles to determine his intent to act. The control system of this mechanical exosuit is able to learn and mimic the natural posture of the wearer. This means that it takes at least 30 minutes for the two to harmonize - one can't expect to be mobile as soon as it's put on. However, HAL-5's primary use is as a rehabilitation therapy aid and nurse's aide, so half an hour of training time doesn't pose a problem. After donning the battery-powered Hercules suit, nurses can easily pick up stout patients as if they were carrying a child. Kayuki Yamakai has begun renting out the HAL-5 to customers. In the Iron Man comics, the superhero is knocked out and lying on the floor of his enemy's lair, with the display inside his helmet informing him of bad news - the electricity is running out. But there is still hope. He plunges his fingers into the concrete floor, finds a wire, and quickly finishes recharging.
Unfortunately, in the real world, exoskeletons are much more cumbersome to charge. As a result, the first XOS on the battlefield might even be connected to a cable. Obusek envisions that this original version was more likely to be an engineer than a warrior. Connected to a power source on a warship or military vehicle, the XOS could help a soldier quickly unload a helicopter loaded with heavy weapons or repair a tank with broken tracks. While the U.S. Army hopes to experiment with a wired version of XOS on the battlefield by 2009, Jacobson and colleagues are still working intensively on a self-powered version.
This summer, Sarkos will work with an engine design firm to develop an engine that can power the XOS for hours on end. Besides, Jacobson wouldn't say much more. He prefers to talk about another, more interesting challenge -- that of reducing the XOS's energy appetite rather than building a powerful engine.
Jacobson showed me a new energy-saving mechanical leg designed to mimic the way a human leg drives. When walking, our hips generate the largest portion of the energy, and when the leg steps forward, the knee and the small muscles in its area relax completely, ensuring that our foot lands in the ideal position on the ground. This free swing technique is quite energy efficient. Kazrouni and Herr have already designed it into their respective lower limb exoskeletons. Jacobson is designing it into a future version of XOS. "The next step," he says, perhaps in a few years, "will be to achieve the goal of walking with one to three horsepower." At that point all that's needed is a portable battery pack to provide lasting power.
Jacobson sees today's version as a base model that will eventually be adapted into various versions to perform different tasks, whether in a hospital or on the battlefield. Future models may even be fully automated. "You walk out and tell it, 'Walk yourself to that building, because I'm too lazy to walk.'"
Later, walking through the lobby of Sakos, I saw flat-screen TVs showing future XOSs depicted in animated cartoons.In one clip, soldiers wearing XOSs carried missiles on their shoulders, flew and jumped over a high wall, and could even do graceful backflips. Though encased in mechanical skeletons, they looked as agile as football sidemen. They look like iron men.