They run, crawl, slither, fly, and jump. They’re also robots. Fueled by funding from the Defense Advanced Research Projects Agency (DARPA) and other public and private organizations, researchers at labs nationwide are developing a new generation of military robots. Inspired by designs already perfected by nature, these robots are helping military units accomplish missions with less risk to soldiers and civilians.
Joseph Ayers, principal investigator of the Biomimetic Underwater Robot Program at Northeastern University’s Marine Science Center, notes that animal physiology and behavior are inspiring robot developers to take military robotics to the next level. “Even the simplest animals outperform any known robot, especially in autonomous operations,” says Ayers. “Animals have performance advantages, and we’re trying to capture these advantages in an engineered solution.”
Ayers’ personal goal is to create a robotic lobster that may someday be used by the U.S. Navy to detonate marine explosives. “The process that a lobster goes through when it searches for food is exactly the same process you would want one to go through to search for underwater mines,” he says.
Ayers isn’t alone in his belief that animalinspired robots are destined to assist or replace soldiers and civilians in a variety of dangerous tasks. “The human is becoming the weakest link,” DARPA warned last year in an unclassified report. “Sustaining and augmenting human performance will have significant impact on Defense missions and systems.”
Yet as researchers strive to create robots that mimic the action of different kinds of realworld creatures, with the goal of removing soldiers from potentially dangerous environments and situations, critics are questioning the basic concept of robot-driven, risk-free warfare.
“The whole field poses a dilemma,” says Noel Sharkey, a professor of artificial intelligence, robotics, and public engagement at the University of Sheffield in the U.K. Sharkey notes that it’s the moral duty of any commander to protect the welfare of his soldiers. “But the problem is, if it’s risk-free war to one side, can that be a just war?”
A nation that feels there’s no risk in engaging an enemy, says Sharkey, may be more inclined to use force to solve its problems. Still, despite the criticism, there’s no sign that the military or any of its funded researchers are feeling any moral qualms about their goals. As Ayers notes, “It’s better to lose a robot than a person.”
Besides being tasty, lobsters are remarkable creatures. They can crawl along the ocean bed almost continuously for up to 100 years, hunting with powerful claws and performing various other activities in places were visibility is often close to zero. Using funds provided by the U.S. Office of Naval Research, Ayers’ submersible robot is designed to use a lobster’s best features to perform dangerous underwater tasks for the military. “It’s designed to eliminate the risk to humans,” he says.
Ayers’ Biomimetic Underwater Robot, unofficially dubbed “RoboLobster,” can move in any direction and wiggle and squirm, just like a real lobster (Fig. 1). Weighing about seven pounds and measuring approximately two feet long, it’s also about the same size as its living counterpart.
Biomimetic robots like RoboLobster are designed to be small, agile, and relatively cheap, particularly when viewed in the light of saving people from injury and death. The systems rely on electronic nervous systems, sensors, and novel actuators. Perhaps most importantly, biomimetic devices take direct advantage of capabilities that have already been proven in animals for dealing with real-world situations in unique environments.
In the case of RoboLobster, the system can even be adapted to improve upon nature. The robot can, for example, be built with “claws” created out of explosives, designed for use in a suicide mine-detonation mission. “Typical mines that are used in underwater warfare are 500-pound aerially dropped bombs,” says Ayers. “No one is going to go in and pick that up and carry that away. All you can do is to detonate it in place.”
RoboLobster is one of the first robots to use artificial muscle. Known as NITINOL, it lets the robot move around easily at depths of up to 40 Naval Ordnance Laboratory) is a family of intermetallic materials that contain a nearly equal mixture of nickel and titanium.
Continue on Page 2
The material’s inherent elasticity and shape memory enables it to bend and snap back to its original position, making it a durable and reliable actuator. RoboLobster moves about on eight legs, each featuring three NITINOLcontrolled motors costing about $200 apiece.
RoboLobster can be powered by a rechargeable nickel-metal-hydride (NiMH) or lithium-ion (Li-ion) battery pack and controlled by a proprietary neuronal-circuit-based controller. But controlling a robot walking out of sight underwater posed a unique challenge for Ayers and his team. To keep RoboLobster from inadvertently wandering away from its target area, researchers devised the idea of using a series of underwater sonar beacons.
“A short baseline array will tell the deviation of that beacon from the orientation of the hull of the robot,” says Ayers. “The idea is to get the robot to home in on the sonar beacon and then swim toward the beacon.”
For Ayers, basing an aquatic robot on a lobster was simply a matter of common sense. “We had an incredible library about how these animals solve problems,” he says. Yet Ayers notes that building a robot based on an animal model is a two-way street. “You need to know how the animal works in order to build the robot. But as you build the robot, you identify things you didn’t know about the animals.”
Ayers plans to begin real-world, underwater testing in about a year. “Everything is built and all the software is working. We’re now in the integration phase,” he says. Ayers believes that RoboLobster can eventually be produced for under $1000 per unit. He feels that price is a bargain, since the Navy is currently willing to spend around $27,000 on equipment and other resources just to take out a single mine. “Plus, you have to consider how much a human life is worth,” he says.
While they aren’t everybody’s favorite animal, rats can persistently pick their way through obstacles to reach whatever they desire, using their highly sensitive whiskers to guide their way. This physical trait has inspired scientists at the University of Sheffield to develop a robot rat. “In designing intelligent, lifelike machines, the use of touch has been largely overlooked,” says Tony Prescott, the project’s leader and a professor of cognitive neuroscience.
Prescott’s rat, currently under development, will be designed to mimic real-world rat behavior. Rats have pretty lousy close-up vision. To compensate, they have evolved a keen sense of touch, primarily through their whiskers.
Prescott notes that the common Norwegian rat uses its whiskers to make sense of its environment. It sweeps its whiskers back and forth at high speeds in a controlled manner, allowing it to use touch signals alone to recognize familiar items, determine the shape and surface of objects, and track and capture prey.
Using their understanding of the animal kingdom, as well as funding from the European Union’s Future Emerging Technologies (FET) program, the Sheffield research team is developing a whiskered robot that can seek out, identify, and track fast-moving target objects. “Overall, our project will bring about a step-change in the understanding of active touch sensing and in the use of whisker-like sensors in intelligent machines,” says Prescott.
Sheffield’s researchers are working on a system that will turn whisker movements into electrical signals (Fig. 2). “There are lots of ways to do this,” says Prescott. “You could use magnetic sensors or strain sensors, for instance.” In either case, the goal is to create a signal that can be processed. “A signal that you can analyze to detect the properties of the surface,” he says.
Prescott envisions several possible applications for the technology, ranging from searchand- rescue robots that could pick their way through rubble and debris, to mine-clearing machines, to planetary rovers in space. The technology could also be used closer to home in domestic products, such as vacuum cleaners that could sense textures for optimal cleaning, he observes.
As a “blue sky project” designed to test a concept, Prescott doesn’t expect the prototype being developed by his team to ever enter production. “We’re looking to see what interest there is from industrial and military users, then develop this technology in the four to five years beyond that,” he says.
SNAKES ON A PLANE AND ELSEWHERE
When people envision futuristic robots, they think of machines that walk, roll, and perhaps even fly. Few, however, would imagine a robot that slithers. Howie Choset, on the other hand, thinks about slithering robots every day.
An associate professor of robotics at Carnegie Mellon University, Choset is working on a robotic snake that could be used for urban search-and-rescue missions. It also could help inspect structures inside aircraft, ships, and other cramped places. And, it could remove land mines.
Funded by the Office of Naval Research and Boeing, Choset aims to create robots that can twist and turn their way into cramped locations that are inaccessible to humans. “With their enhanced flexibility and ‘reach’ ability in convoluted environments, serpentine robots make sense,” he says. Snake robotics has been a career-long venture for Choset. He embraced the idea as a student in 1971. This project began in 1990.
Continue on Page 3
Choset notes that robotic snakes would be perfect for investigating destroyed structures. Large buildings that experience a pancake collapse, such as the World Trade Center, prevent rescue workers from entering due to fear of further collapse. “But the biggest problem is that both people and dogs are usually too big to enter the voids between debris,” he says.
The technology could also be used to check the condition of wounded soldiers in a battlefield environment. “It’s a shame to have a soldier go out and check to see if someone is hurt or dead in the middle of a firefight,” says Choset. “One of the most common causes of casualties in the Iraq War is someone going out to help a solider they think is hurt.”
A low-profile robotic snake, equipped with video and audio sensors, could assess the situation without placing additional lives at risk. Choset also envisions robot snakes slithering up flagpoles and antenna masts to provide onthe- spot reconnaissance.
Biological snakes move by using different cyclic forms of squirming locomotion, or “gaits.” Adapting these gaits to the mechanical snake enables the robot to maneuver through 3D terrains. Choset says his specially designed motion-planning algorithms allow his electromechanical snakes to autonomously sense and respond to everything they encounter.
Choset explains that tips from evolution helped him design the snake robot. He points out that snakes lost their legs because they got in the way when crawling through narrow passageways. That’s why he decided to use squirming locomotion for a robot designed to enter tight locations. Using beveled gears around its circumference, Choset’s robotic snake—measuring only 5 cm in diameter—has many more degrees of motion freedom than just about any other robot ever developed (Fig. 3). Built-in redundancy allows the robot’s mission to proceed even if an actuator or other key component fails.
Choset admits that several significant challenges still must be addressed before robotic snakes can go into full production. Mechanism design, control, and sensor integration are some of the remaining major issues. Better path planning is also needed, he says.
“Current serpentine path planners are rudimentary at best and can’t work in complex environments,” says Choset. He’s currently developing new algorithms for directing serpentine mechanisms through unknown 3D spaces. “It’s very complex work.”
CAUGHT IN A SWARM
Insects may be an even more exotic model. Thanks to their small size and ability to work in unison, robots designed on insect principles may be able to accomplish tasks faster and more efficiently than much larger machines.
The goal of the European Union-funded Symbiotic Evolutionary Robot Organisms project, also known as Symbrion, is to understand the principles that govern how robots can form themselves into a single artificial organism. This approach allows robot “swarms” to interact collectively with the physical world (Fig. 4). The technique could ultimately be applied to realworld tasks, such as finding and reporting on the location of wounded soldiers and civilians.
Multi-robot organism swarms would be designed to contain anywhere from a dozen or so to hundreds or even thousands individual robots. Each of the mobile (but not airborne) devices would be only slightly larger than a sugar cube, but capable of working together as a single artificial lifeform.
The robots would be able to share information and energy with each other as well as manage their own hardware and software. In fact, when the devices join together into a single organism, each will be able to share its crucial information with all of the others, creating an overall system that can evolve in the face of new problems—just as a natural immune system can cope with unfamiliar pathogens.
Jon Timmis, a researcher in the Department of Electronics at the University of York in the U.K., is working to develop an artificial immune system that would protect individual robots and the larger collective organism. The immune system, says Timmis, will be able to detect faults and make recommendations to a high-level control system about corrective action, much in the way a person’s adaptive immune system monitors the body’s status to keep it healthy.
The project’s ultimate goal is to create autonomous swarms that can work both separately and as a group to accomplish specific tasks. “You might have 500 individual units running around doing individual tasks and then, under certain conditions, they will join into a single robotic unit,” says Timmis.
For example, if a military unit is reportedly lost, multiple robots can set out on a search mission. Then, once the missing soldiers are found, the swarm will join together to deliver supplies, provide airborne reconnaissance, and perhaps even carry the most seriously wounded personnel to safety. “This swarm of robots will be able to figure out, on their own, that they need to join together, as well as figuring out how to do this, in order to achieve their task,” Timmis says.
Timmis believes the first swarming robots should begin arriving in just a few years, although many more years of R&D will be required to create systems capable of searchand- rescue and other useful activities. “The core technologies will be in place in five years, and there will be demonstrations of robotic units joining together, climbing over walls, and so on,” he says. “I’m very confident this will happen.”
Continue on Page 4
The military’s current infatuation with animalinspired robots resembles previous attempts to use trained animals as helpers and even weapons. The U.S. Navy, for example, used dolphins to help locate and clear mines in the Persian Gulf during the 2003 Iraq War, and it continues to train dolphins and sea lions for military tasks at West Coast facilities.
Sharkey notes that today’s robot experimenters are now, in essence, picking up the ball and heading toward the same goal as previous animal experimenters. Therefore, many of the same moral issues animal researchers faced are now being tossed at military robot developers, particularly as robots move from support roles to fighting enemy combatants.
“No matter what physical form the robots may take, the drive is to have fully autonomous robots that can decide for themselves who to kill,” Sharkey says. “That’s a serious ethical challenge.”
Besides the moral dilemma posed by autonomous fighting robots, Sharkey worries the technology may eventually fall into the hands of unstable regimes and terrorists—people who would feel no compunction about sending murderous robots rolling, crawling, or slithering toward innocent civilians. “Using robots doesn’t mean you won’t have suicide bombers as well,” he notes. “It just adds more to the arsenal.” For now, most animal robot researchers say they are content to build machines that are designed to save human lives rather than take them. As Ayers puts it, “It’s kind of hard to imagine a fighting lobster.”