TMS Logo  TMS ONLINE | TMS PUBLICATIONS | SITE MAP

An Article from the November 2003 JOM: A Hypertext-Enhanced Article

The author of this article is managing editor of JOM.
Exploring traditional, innovative, and revolutionary issues in the minerals, metals, and materials fields.
OUR LATEST ISSUE
READ JOM ON-LINE

cover

VISIT THE
JOM COVER GALLERY

JOM MENU
 
 
Feature: Materials World

Personification: The Materials Science and Engineering of Humanoid Robots

Maureen Byko


Figure 1

Figure 1. ASIMO’S climbing techniques are a first for humanoid robots. Photo courtesy of Honda Motor Company.

INTRODUCTION

Contrary to the musings of pulp science fiction writers, robots pose little threat of taking over the world anytime soon. They can barely walk, are incapable of independent thought, and run out of energy quickly.

In laboratories around the world, though, researchers are working at solving those problems and others, opening the door for development of a robot that can mimic humans. Progress is evident in several arenas. For instance, within the last year robots have begun to populate nursing homes, where doctors use them as a “remote presence” to monitor patients. In the last two years, a major international robot competition added a humanoid category, with the ultimate goal of a robot-human competition. One company has developed a humanoid that it suggests could provide friendly conversation based on its owner’s interests. Maybe world domination is not in the future of humanoid robots. But making the world a more interesting place just might be.

SAY HELLO TO ASIMO

It rang the opening bell on the New York Stock Exchange. It greeted visitors to Honda Motor Company’s headquarters in Japan. It shook hands with the prime minister of Thailand. This well-traveled mass of magnesium is a humanoid robot known as ASIMO, an acronym for Advanced Step in Innovative Mobility (Figure 1). As hinted by its name, the robot’s claim to fame is its ability to walk and climb stairs. ASIMO, billed by Honda as the world’s most advanced humanoid robot, is the result of 17 years and untold millions of dollars in R&D.

With a body of lightweight magnesium coated with a plastic resin, ASIMO was an exercise in mobility research, said Jeffrey Smith, leader of the ASIMO project in North America. The intent, Smith said, was to design a robot that was mobile, able to maneuver in household environments, friendly to look at—cute, even—and tall enough to look into the face of a seated human to receive voice and visual messages, such as “open the door” or “turn off the light.” “On the most basic level, Honda built ASIMO to help people,” Smith said.

The work began in 1986, with a two-legged robot that walked very slowly and fell frequently. Subsequent iterations were based on exhaustive research into the mechanics of walking of humans and animals. The mobility of the current design benefits from what is called “prediction motion control,” which allows the robot to predict its next move and shift its center of gravity accordingly. Earlier versions walked and turned according to combinations of stored walking patterns rather than making real-time adjustments. In addition, ASIMO has 26 degrees of freedom, including movable knees, hips, and feet, all of which help the robot maintain control while walking.

One day, the company envisions the robot helping the elderly or sick, but the cost will have to come down significantly first. Not counting the R&D investment, ASIMO cost about $1 million to build, Smith said, and inexpensive mass production is still far enough in the future that no one from Honda will speculate publicly when that might occur. Power is another obstacle to practical use. The robot is powered by a 40 V nickel metal hydride battery, and can operate approximately 30 minutes before the battery must be recharged. Although lithium-ion batteries might offer improved performance, “the future of ASIMO is in fuel cell power,” Smith said. “That’s a long way off.”

The robot, while equipped with voice and face recognition technology, is controlled by wireless computers and was not designed to be an artificial intelligence showcase, Smith said. “Its real intelligence is in the advanced technology with which it is equipped,” he said. Still, when the robot, currently on tour in the United States, enters an auditorium and walks smoothly to the edge of the stage, surveys the crowd, and waves both hands with a cheerful “Hello everyone,” there is an appearance of personality. That friendly, human-like appearance, Honda officials believe, will make the robot more acceptable for service to humans.

ARTIFICIAL MUSCLES MAKE THEIR MOVE

Even the most human of machines still move in a mechanical way, accompanied by the noise of a motor. The answer to quiet, human-like movement could lie in electroactive polymers used as artificial muscles. Electroactive polymers (EAPs) are plastics that move in response to electricity. They are the reason robotic fish swam in a fluid, natural way when they were displayed by EAMEX, of Japan, at the International Society for Optical Engineering 2003 EAP Actuators & Devices (EAPAD) Conference. And they are the reason the world may be treated to an arm-wrestling match that pits a robotic arm against a human in 2005.

Ron Pelrine, program director of the advanced transducer program at SRI International, a leader in EAP development, described EAP as a rubbery polymer made in sheets. Electrodes, which must be able to stretch with the material, are placed on each side of the sheet. A voltage is applied across the electrodes, causing the electrostatic forces from the plus charges and minus charges to squeeze the polymer, causing it to stretch out, creating forces that Pelrine called “pretty substantial.” With these materials beginning to move out of the realm of experimental and showing promise for practical applications, interest is high from many areas, including robotics, Pelrine said. “ There’s a lot of interest these days in more life-like robots, and not necessarily just from an aesthetic point of view, although the entertainment industry is certainly a big one for robotics, but also in terms of the ability to move over obstacles, the ability to move quickly.”

This kind of progress is exactly what Yoseph Bar-Cohen was hoping for when he proposed the arm-wrestling contest in 1999. Bar-Cohen, a physicist and senior research scientist at NASA’s Jet Propulsion Laboratory in Pasadena, California, and organizer of the first EAPAD conference in 1999, decided that competition is often a successful means of stimulating R&D (see sidebar on humanoid competition). So he posed the challenge that requires the robotic arm to be EAP-actuated. The response came much quicker than expected.

“I thought 20, 30 years from now someone would do it,” Bar-Cohen said. “It would probably be when I was retired and I would watch from the side to see how things develop. I didn’t realize that five years later someone would come and say ‘I can do it.’”

That “someone” is SRI. The only obstacle to SRI’s participation is funding, said Philip von Guggenberg, director of business development.

“Technically, we can meet the challenge,” he said. And by 2005, not the 20 years or more Bar-Cohen expected.

Because the company is a non-profit organization, it cannot cover the cost of participating the contest on its own. Von Guggenberg is hoping the contest will award a monetary prize that will compensate for the cost of participating. Bar-Cohen is now considering how to find sponsorship of such a prize.

THE DOCTOR IS "IN"


Figure 2

Figure 2. A doctor consults with her patient using a “Companion” robot. Photo courtesy of InTouch Health.

It’s low-tech, with no legs that climb or artificial muscles for smooth performance, but one robot is proving successful as a substitute for humans. InTouch Health, based in California, developed robots that can be controlled from remote locations through a broadband and wireless network. Instead of a head, the robot’s body is topped with a video screen onto which the controller’s face is projected (Figure 2). The robots are controlled mainly by doctors who can see their patients even when they are not in the same building. Using joysticks, the doctor can move the robot around the patient. Using the zoom lens, the doctor can visually examine the patient. Using auditory transmissions, the doctor can have a natural conversation with the patient. The patient, looking at the computer monitor, can see the doctor speaking.

“What we’re all about is allowing someone with some medical expertise, who can be located anywhere in the world, to extend their presence to any other location where their knowledge base is valuable,” said Tim Wright, vice president of marketing for InTouch. InTouch was established in 2002 to develop a robot to fill the labor gap in the health-care industry. Statistics prove a current shortage of health-care workers will only grow greater with the aging of the baby-boom generation (birth dates 1943–1960 in the United States). According to a May report by the U.S. Department of Health and Human Services, the demand for direct-care workers will grow to 5.7–6.6 million workers in 2050, an increase of 200 percent and 242 percent, respectively, over the current 3.8–4.6 million demand. As of October, the InTouch robotic doctors were in use in four United States nursing homes and two hospitals. The design may not have ASIMO’s advanced technology, but neither does it have that robot’s price tag. The medical robots, called “Companion” devices, lease for $3,000 a month.

The robots, made of aluminum, are slightly short by human standards, at 1.6 meters tall, and a little overweight, at 98 kg. They have no arms or feet, but move on three urethane balls that allow them to roll easily in all directions. The robots have sensors that alert them to the presence of people and objects in their vicinity. Wright said the company’s technology teams are working on arms now, so the robot could be capable of such tasks as pushing wheelchairs or opening elevator doors.

Research is continuing, Wright said, not only into arm development, but also into humanizing the shape. The robot is now little more than a vertical, rectangular box floating on a wide base. The next versions will likely have more of a waist and shoulders, a change based on customer comments that they would be more comfortable with a human-like machine. “We’re looking for a look that is professional and also friendly,”Wright said.

THEY GO IN PEACE



Figure 3
Figure 4

Figure 3. NASA’s Robonaut could keep human astronauts out of harm’s way. Photo courtesy of NASA.

Figure 4. Robonaut working with astronaut Nancy Currie to build an aluminum truss. Photo courtesy of NASA.


NASA, too, is exploring the benefits of supplementing its workforce with robots. The Robonaut, a humanoid robot operated with a telepresence control system, is considered a workable candidate for space station repairs (Figure 3). The robot is being developed by the Robot Systems Technology Branch at the Johnson Space Center in cooperation with the Defense Advanced Research Projects Agency. It consists of a head, a torso, two arms, and two five-fingered hands designed with human-like dexterity. Robonauts, NASA reports, could work alongside humans or in environments considered too risky for people.

To ensure that the robot can withstand the stresses of space, its endoskeleton is made up of gold-anodized aluminum alloy parts. The forearms and palms are cast, due to their complexity, and then machined to specified tolerances. The torso of the robot contains the system’s central processing unit, a large electronic junction board, distributed power converters, and numerous wires and connectors. Those are protected by the endoskeleton’s rigid carbon fiber breastplate and backpack. Two types of helmets are being tested: one of a resin that is hardened in a sterolithography process (as shown in Figure 3) and the other made of sintered glass fibers.

The Robonaut design was tested this summer in working with human teams to assemble an aluminum truss structure (Figure 4). Astronaut Nancy Currie, who participated in the test, sees potential benefits to working with Robonauts on the International Space Station. “On the station, you could send a Robonaut or two out early to set up the work site, or leave them out late to clean up,” she said. Normally two astronauts would be doing those tasks. The Robonauts are expected to be ready for work on the space station in three or four years, according to test conductor Robert Ambrose.

AN AGREEABLE COMPANION

Robots have been working for humans for years, in one form or another. One day, their purpose might simply be to play with humans, or at least entertain them.

In “Robbie,” a 1940 short story by science-fiction writer Isaac Asimov, a young girl named Gloria has a robot, Robbie, programmed to be her caretaker. Gloria’s love for that machine concerns her parents. Their daughter does not understand that this machine is just a mechanized playmate, and that blending of machine and human emotion is unhealthy, her parents believe.


Figure 5

Figure 5. QRIO robots by Sony were unveiled in October. Photo courtesy of Sony Corporation.

The story was set in 1998. In October 2003, Sony Corporation unveiled QRIO (Figure 5), a potential playmate for modern-day Glorias. This humanoid creation is described on the Sony web site (sony.net/SonyInfo/QRIO/story/ index_nf.html) as “the product of cutting edge artificial intelligence and dynamics technology.” Satoshi Amagai, president of the Sony Entertainment Robot Company, even suggested in an interview that the mission of QRIO is “to bring people emotional value.” “We want to create something that’s fun to be with, can be of help at times, even something that one can talk to when one is lonely,” Amagai said.

Like ASIMO, the robot is highly mobile. Made of magnesium, it is shorter than ASIMO (half a meter), weighs less (7 kg), and has 36 degrees of freedom. QRIO is capable of fluid movement due to Honda’s Intelligent Servo Actuator, an integrated actuating system that controls the robot’s joints and limbs, allowing instant reactions to outside influences such as uneven walking surfaces or a push on the robot’s body. It is equipped with two cameras in its head and can recognize faces. Likewise, the robot has seven built-in microphones for voice recognition. Autonomous behavior control architecture allows QRIO to react to its environment as it moves, and Sony claims, to display emotion based on external influences.

At MIT, a professor has made it her mission to instill the illusion of emotion in robots. Cynthia Breazeal, with a doctorate degree in electrical engineering, is director of MIT’s Media Lab Robotic Life Group. As her doctoral thesis, she developed a robotic head she called Kismet (recently retired to the MIT robot museum) that was capable of expressing a wide range of emotions such as surprise, fear, and pleasure. As shown on the cover of this issue, the robot had lips, eyes, eyebrows, ears, and a neck, all of which were programmed to move in certain ways to express emotion during human interaction. In addition, Kismet could speak, but only gibberish, with appropriate inflections.“We can actually have the robot sound angry when it’s angry, sound sad when it’s sad, and so forth,” Breazeal said.

Breazeal has lately turned her attention to high-level toy making. She is working with the Hollywood special effects company Stan Winston Studio on a furry creature called Leonardo. Leonardo, like Kismet, is able to express emotion, but it also has arms, a torso, legs, and skin. The face is designed with 32 degrees of freedom to allow for minute, lifelike changes of expression. The skin is designed with embedded sensors that allow it to respond to touch. Pet its ear and the ear twitches. Grab its hand and it pulls back.

The sensitive skin, the emotional eyes, the artificial concern all might one day mimic humans too well. Bar-Cohen, who in May co-authored a book with Breazeal called Biologically Inspired Intelligent Robots, said one area of concern is how people would react to a lifelike robot.

“When you look at some of the movies and what the robots would do . . . we are very uncomfortable because they could beat us,” Bar-Cohen said. “Computers are already beating us in capability now; imagine if they all look like us.”

FAVORITE FICTIONAL ROBOTS OF THE ROBOTMAN



Figure 6
Figure 7

Figure 6. Robotman Fred Barton and three of his re-creations: from left, B-9 from Lost in Space, Gort from The Day the Earth Stood Still, and Robby, from Forbidden Planet. Photo courtesy of Fred Barton Productions.

Figure 7. Another humanoid favorite, C-3PO from Star Wars. ©Lucasfilm & TM. All rights reserved.


Fred Barton, also known as the Robotman, makes his living making robots (Figure 6). He built his first robot in high school—a full-sized replica of Robby, the robot from the classic 1956 science fiction movie Forbidden Planet. He went to college and attempted more conventional careers, but always dabbled in robots. Now, they are his livelihood, as he builds replicas of popular robots that sell for anywhere from $5,000 to $40,000. Who better, then, to help JOM compile a list of noteworthy robots of the large and small screen?

  1. Maria (1926): The movie Metropolis featured a female robot, a rarity in film. In the movie, Maria is created when the soul of a human woman is transferred to the female robot, and an evil robot is the result. “They were always made the heavy,” Barton said of the early science fiction robots.
  2. Gort (1951): The robot Gort appeared in the movie The Day The Earth Stood Still to send humans a message about the dangers of atomic weapons. Gort is his second most frequently requested replica, Barton said.
  3. Robby (1956): Forbidden Planet introduces Robby, an intelligent, well-mannered robot. Robby is Barton’s favorite and the most-requested replica. “Robby raised the bar and made science fiction more respectable and less silly.” Robby, who later appeared in numerous other movies and television, can be seen in a movie to be released this month, Looney Tunes, Back in Action. Barton’s replicas appear in the movie along with the original Robby, which he restored.
  4. B-9 (1966): In the television series Lost in Space, a robot designed by Robby’s creator is a regular, living with the Robinson family on a remote planet. “He was more or less Will’s camp counselor, always under orders from Will’s dad to protect him and keep him out of trouble. He was under Dr. Smith’s evil control at first, then broke free of that.”
  5. C-3PO (1977): This robot (Figure 7), along with its sidekick, R2-D2, appeared in Star Wars. Lucasfilm describes the robot as “fussy and worry-prone.” Barton interprets that as an effeminate personality. “ Usually they don’t assign genders to robots.”


CYBORG: COMBINING MAN AND MACHINE
Perhaps the best sign that Kevin Warwick’s work has merit is in his mailbox. “I get quite a few letters from people with all sorts of different problems . . . a lot of people that have paralysis or partial paralysis,” he said. Warwick, professor of cybernetics at the University of Reading in the United Kingdom, has twice had mechanical devices surgically implanted in his body so he could operate as part man, part machine. He calls himself the world’s first human cyborg. “This (research) is looking at upgrading humans, giving humans extra abilities,” Warwick said.

Warwick’s most recent surgery, in March 2002, implanted a 100-electrode array into the median nerve fibers of his arm. The implant remained in place for three months. Previously, in August 1998, he had a silicon chip transponder implanted into his forearm. The first implant allowed a computer to monitor Warwick. The chip emitted a unique signal, which was received by a computer. That computer was programmed to perform such tasks as opening doors and turning on lights when it sensed Warwick’s presence. The 2002 neural implant, made of silicon, platinum, and titanium, communicated with a radio transmitter and receiver that were linked to a processing unit (Figure A). During the experiment, Warwick succeeded in transmitting signals to control an electric wheelchair and an intelligent artificial hand using the neural interface. He also received signals from his wife, who volunteered to have a transmitter implanted in her arm as well. The goal of the experiments was twofold: to see if the implants could be used to help people with paralysis, or who have lost control of particular body functions, and to test whether such implants would be rejected by the body. The conclusions on both counts were positive, Warwick said.

“My body, far from rejecting the implant, it actually seemed both mentally and physically to link with it more and more as time went on. In fact, it was difficult for the surgeon to remove it in the end because body tissues grew around it and held it closer to the nerves,” he said.

He described the communication that went on with the implant as “misusing movement signals. . . . I could move my fingers around and we used the neural signals that control that to move a wheelchair around.” Warwick said he would like his next experiment to be an implant in his brain.

In fact, brain implants were reported to be successful in monkeys in a paper published in October in The Public Library of Science. In those experiments, Duke University researchers implanted microwires into several brain regions of two monkeys. Next, the monkeys learned to move a joystick that controlled a robotic arm, which appeared on a computer screen. Once that skill was mastered, the researchers disconnected the joysticks, and eventually, the monkeys were able to move the arm with only neural signals that were interpreted by a brain-machine interface as they watched their success on the computer. With time, they improved their ability to manipulate the robotic arm without moving. The success is considered promising to eventually help humans with paralysis bypass muscles that are not functioning using similar brain-machine interfaces.


Figure A

Figure A. The device that allowed Kevin Warwick to transmit and receive neural signals. Photo courtesy of iCube Solutions.

 


COMPETITION PUSHES HUMANOID PROGRESS
Every year, robots and their operators gather from around the world to show their skills, including soccer capability, at RoboCup. In 2002, a humanoid league was added to the event, and Changjiu Zhou, a professor in the School of Electrical and Electronic Engineering at Singapore Polytechnic, was named as co-chair of the Humanoid League. Zhou leads a research group working on intelligent robotic systems in general and humanoid robots (Robo-Erectus) and PDA-based educational robots (Robo-Tutor) in particular. The Robo-Erectus team placed first in the free-style competition this summer at RoboCup 2003. In this interview, Zhou shares some thoughts on RoboCup and humanoid robots.

Q. What is the technical goal of creating a humanoid robot that plays soccer?
A. The long-term scientific goal of the RoboCup is to develop humanoid teams that could beat the World Cup champions by 2050. A humanoid robot, because of the intended similarity of its physical structure and locomotion to human beings, is the most ideal robot for mankind. A humanoid robot should be able to perform all manual or even mental tasks as well as, if not better than, a human being. It should be able to navigate freely in the environment that the human lives in, communicate effectively with humans using natural languages, and respond appropriately to various life situations of human beings. Soccer is a game that best illustrates the various complex skills of a human. These include locomotive skills like walking, running, kicking, jumping, picking up a ball, etc.; perceptive skills like recognizing the terrain and identifying the ball, goal post, team mates and opponents; mental skills like role playing, path decision, tactics, and strategy; and communication skills either linguistic or in gestures, in addition to group behavior skills such as coordination, passing, deceiving opponents, attacking, and defending. By setting the target of a soccer game between human and humanoid robot, it is hoped that all these human skills could be replicated successfully in a robot.

Q. Is there also a benefit to such humanoid competition because it attracts public attention and possibly raises awareness of robot research?
A. Yes. The main goal of humanoid competition is to stimulate research and development in this area. The technological advancement that has been made year by year is very observable. It definitely attracts public attention and even prompted some private individuals to make their own humanoid robots to take part in the competition.

Q. Do the teams share information or learn from one another?
A. Yes, the teams do somehow share information to learn from one another. But what I’m seeing of sharing is far from enough. For example, there is currently so little information available about how to develop humanoid robots. Typically, researchers have to build robots from the ground up. They have to assemble motors, sensors, and other parts, and put an operating system and computing intelligence into the robot before they can begin their research and prepare for robot competition. Therefore, sharing across leagues and between teams is very important. Beating the World Cup champs definitely requires worldwide cooperation. On the other hand, RoboCup creates a sharing atmosphere. By bringing people that are involved in robotic R&D from all over the world together, the competition is definitely a meeting place and a forum for them to share experiences, to show and also to witness technical advances made by all teams, and of course through interaction whereby they are together in the same competition venue for a good number of days, the participants do learn from one another. In addition, a symposium is organized in conjunction with the competition to allow the participants to present technical papers on their achievements.

Q. With so many people working on so many aspects of a humanoid robot, such as walking, facial expressions, and vision, is there a chance that everyone will combine their knowledge anytime soon?
A. Good question! Integration is one of the biggest challenges in the field of humanoid robotics. Whilst it is not that difficult to build a vision system or control mobility, plan obstacle-free paths, localize or manipulate objects, it is hard to do all these things at the same time with both high reliability and secure recovery procedures in the event of a subsystem failure. In the domain of humanoid robotics the integration stage cannot be avoided and we are forced to face the challenge front on and with new methodological approaches which go beyond simply aiming to perfect each component. At the present moment, we are seeing the integration of vision and locomotion control; in times to come, other aspects like facial expression and natural language will be integrated.

Q. How have the advancements in computer power benefited robot research?
A. As humanoid vision, natural language processing, and on-line gait generating are computation-intensive, advancements in computer power will definitely make the realization of a humanoid robot closer. From a practical standpoint, robots require a lot of computing power to perform their jobs. For example, the ASIMO robot that Honda is developing has five Pentium III Processor-M processors inside, all performing different functions. So we see robots representing a very significant market for processing power—not in the next couple of years, but in the longer term.

Q. In movies and books, people have imagined robots so intelligent they could take over the world. Do you still find people who have such fears about your work? Or do they trust that humans will always be superior?
A. I don’t believe for a moment that robots will dominate the earth. Robots can either be very useful—but then they are specialized and usually not very autonomous or cognitive—or they can be entertaining, like pet robots and toy robots. On the other hand, a humanoid robot, no matter how smart or intelligent it may be, is still a man-made and programmed machine and not a living organism. Unless we were able to program all our capabilities into a robot, there is no fear of robots posing a danger to human beings.

 


Copyright held by The Minerals, Metals & Materials Society, 2003

Direct questions about this or any other JOM page to jom@tms.org.

Search TMS Document Center Subscriptions Other Hypertext Articles JOM TMS OnLine