From Science Fiction to Reality: Personal Robots

Posted on: Jun 15, 2007

Professor Wayne Book and colleagues at four other universities are designing the Compact Rescue Crawler robot. Book's lab is contributing its expertise on haptics -- the 'feel' associated with operating a mechanical device.

Georgia Tech Photo: Gary Meek

It's 6 a.m., and the Clarks awake to fresh coffee served to them by Millie, one of the family's personal robots. As they get ready for work, Millie makes the bed, and their robotic dog Mickey gently reminds Mr. Clark to take his medicine.

Once at work, Mrs. Clark, a hospital nurse, assigns a personal robot to deliver blood samples to the lab while she talks with a patient. Meanwhile, Mr. Clark catches the morning news while his autonomous car navigates the traffic into the city.

At day's end, the family returns to a spotlessly clean home courtesy of Millie's untiring work. The Clark children do math homework with tutoring from Margie, another robot. After a dinner the Clarks prepared based on a menu suggested by Millie, the family enjoys the rest of the evening free from chores. They sleep soundly knowing that Mickey is always alert to any trouble.

This scenario is not a page from a lost “Jetsons”script. It's likely to be a normal day in the life of a family in as few as 20 years from now, according to robotics experts at the Georgia Institute of Technology.

Already, the global market for personal robots is growing 400 percent a year, says Professor Henrik Christensen, director of the newly formed Robotics and Intelligent Machines Center in the Georgia Tech College of Computing.

“Personal robots are becoming more popular as people want to do more and more with their lives,” Christensen says. “Technology is making it possible...We live stressful lives now, and we can use technology to take away the boring parts of everyday life.”

Robots are not novel technology in industry, the military and even space exploration. But a new generation of intelligent machines called personal robots -- ones that work with and directly for humans, especially in the home, workplace and school -- have begun to emerge only recently. A confluence of smart materials, low-cost, high-speed computing power, better batteries and knowledge of how humans interact with machines is creating an explosion in the market for personal robots, researchers say.

“To have a personal robot that does things you need, you have to have onboard processing, perception, motion and power,”says roboticist Tucker Balch, an associate professor in the College of Computing.“Until two or three years ago, you couldn't put all of that on one small, light platform. Motors and computers take a lot of energy, and the batteries we had couldn't do the job.

“Now, demand for better cell phone and laptop batteries is driving improvements,” Balch adds. “Until recently, you couldn't get enough processing power without drawing lots of electricity. Also, robots on the market now have addressed the high power requirements of motors. Finally, we have all the technologies that can support a consumer robot that is not too expensive.”

Balch predicts that truly useful, multi-function personal robots will cost between $1,000 and $1,500. Single-purpose robots, such as the Roomba vacuum cleaner already on the market, cost between $150 and $300.

While some personal robots are already available, important research is under way to address the remaining technical and societal challenges. Georgia Tech researchers in computer science, engineering, psychology and the liberal arts are collaborating under the umbrella of the new Robotics and Intelligent Machines Center that Christensen directs. That cooperation is vital to creating the best-designed personal robots.

“If you just have computer scientists designing robots, you're not going to build a robot that's as good as one that could be built by computer scientists and mechanical engineers working together,” Christensen says. “We are leveraging Georgia Tech's world-class expertise in all of these domains and want to make something that no one else in the United States is doing today.”

The center's research agenda draws upon Georgia Tech's long tradition of robotics study, as well as findings from an ongoing analysis of 40 Georgia companies. Christensen and Professor Steven Danyluk, who heads Georgia Tech's Manufacturing Research Center, are identifying the problems preventing companies from integrating robots into their operations.

Solving industry and workplace problems -- such as robotic robustness and perception -- will lead to better robots in the home and school, researchers say.

“In our lifetimes, we will have a Rosie (of 'Jetsons' fame), the ultimate home assistant,”Christensen says.

Technical Challenges

Before personal robots become part of daily life, improvements are needed in personal robot software, robustness, materials, component integration, power and human-machine interaction, researchers say.

“Two key chunks of missing technology are perception and reliability, and research is focusing deeply on these issues,” Balch says.

Perception involves the processing of information from a robot's sensors so the robot understands the outside world — at least enough to know what it should be doing.

“Robots need to be able to interpret their world,” Christensen explains. “If they go in a new environment, they need to be able to recognize, for example, a chair even though it's a different chair from one they've seen before.”

From a reliability standpoint, the robot needs to be able to realize when it's stuck and call for help. “Even better would be that it not get stuck or that it can get itself unstuck,” Balch says.

Personal robots must be robust, able to function 24 hours a day, 7 days a week in a variety of environments. Their industrial counterparts already are being asked to work in an array of environments, including temperature extremes ranging from freezing to 100 degrees or more. Emerging industrial application areas, including poultry processing, require units to work 16 hours a day and also endure a daily cleanup process that employs high-pressure
water and caustic chemicals.

“Designing a robot to survive in this environment is difficult,” says Gary McMurray, a senior research engineer in the Georgia Tech Research Institute (GTRI). “You have to protect the electronics and sensors, so material selection is important. We'll have to move away from lubricant use for robot joints, and we'll need the right types of motors and drive systems.”

Materials used to build robots must not only protect components, but also protect the humans that interact with the machines. That requires the development of flexible materials, Christensen says.

For example, robotic arms need to be as flexible as the human arm, which won't break easily, yet as stiff as the human arm when it lifts and pushes, he explains. An example is a lightweight robot that naturally yields when pushed upon; it is based upon Georgia Tech research and manufactured by the Atlanta company CAMotion Inc.

Another technical challenge is the integration of various products into one robotic system. Microsoft is attempting to address this problem with its new Robotics Studio operating system, though it will face competition from other companies vying to create the robotics operating system of choice, Christensen says.

Balch predicts that a standard operating system will accelerate robotics development like IBM's PC did in the early 1980s. “Microsoft is now helping define a standard that's not been there, and I think that companies waiting to enter the robotics marketplace now will enter it,” Balch says. “Combined with the hardware that's available, this will be the last domino to fall.”

If component integration is the final piece of the puzzle, issues of robot power and human interaction must be addressed first. Better batteries might allow robots to operate untethered for long periods of time, says robotics expert Wayne Book, a Georgia Tech professor of mechanical engineering.

But current batteries are way below the necessary levels of operation. Alternatives are being studied in the Center for Compact and Efficient Fluid Power funded by the National Science Foundation. In building a robot called the Compact Rescue Crawler, Book and his colleagues at Vanderbilt University are addressing the power issue by using energy generated by chemical fluids called monopropellants, such as hydrogen peroxide.

In human-robot interaction, hurdles remain in ease-of-use and communication. Balch likens the goal for personal robot ease-of-use to the simplicity of the TiVo digital video recorder interface. “It is a technology that you can give to a 70-year-old and not have to worry about helping her with it,” Balch says.

For humans to effectively communicate with personal robots, the machines need to be able to understand spoken language and gestures, Christensen says. For now, those capabilities are limited.

“The big question is, 'How can people tell a robot what they want it to do?'” Balch says. “People need to be able to show their robot how to do something. Researchers have lots of ideas on how to do this, but the problem is not solved yet.”

One researcher in GTRI is seeking insight by focusing on opportunistic human-robotic interactions that will enable people to work with robots, rather than commanding them.

Researcher Lora Weiss is analyzing both social and mathematical networks to understand the dynamics of robot-to-robot and robot-to-human interactions. She is studying these relationships via software and assessing the larger network of dynamic interaction. Her goal is to mathematically capture how humans behave toward machines from a systems perspective.

“Within software, you can provide some intelligent automation to the bots, and then have a system of real people interacting with the machines,” Weiss says. “The software approach allows one to rapidly populate scenarios with disparate entities and evaluate the emerging and evolving behaviors of the larger system.”

Social Acceptance

Robots running amok has often been a theme of science fiction. “One of our biggest enemies is Hollywood,” Christensen says. “The view of robots that Hollywood projects is almost always negative.”

Christensen believes the public's concern about robots running amok is unrealistic because technology developers place so much emphasis on safety.

“We have to overcome misconceptions about robots,” Christensen says. “We cannot afford one failure….We need to make robots that are cute and fun and interact socially with people, but actually help them in their everyday lives.”

In the workplace, for example, Book says convincing people that a lightweight robot can work safely hand in hand with humans, while also being durable and effective, is a more significant challenge than the technical issues.

“Once we can overcome the perception that lightweight robots are flimsy, then every industry will be happy to save money by using these robots,” Book says. “When industry starts to accept that lightweight robots can do the job, the perception problem will become a non-issue.”

To build robots that people will accept and even like, researchers draw upon studies in psychology and human-computer interaction.

“There is a general hypothesis that robots similar in appearance to what is familiar to us will ease acceptance,” Balch notes. “If you see a robot that looks and acts like a puppy, you're going to treat it as somewhat of a subordinate, but gently. You will guide it.”

On the other end of the spectrum, there are human-looking robots. “You then expect it can do things like a person, and you're less patient with it,” Balch says.

That raises the question of whether robots should look like humans. “The more a robot looks like a person -- it's called the uncanny valley theory -- the creepier it seems to humans,” Balch explains. “If you could design a robot that truly looks like a person, people might accept it, but if it's off target, it's creepy.”

What the Future Holds

Challenges remain for researchers and society in assimilating personal robots into everyday life.

Opportunities exist for business, industry, schools and our lives at home. Also, questions of ethics arise, and people are likely to wrestle with these issues.

For business and industry, robots are another technology that helps countries compete in the global marketplace, says Craig Wyvill, chief of GTRI's Food Processing Technology Division. “While industrial robotic technology continues to evolve, it is already demonstrating it can increase product quality and help companies establish themselves as leaders in their fields,” he explains. “In the process, the workforce must change to support the technology. These changes are essential to staying competitive.”

In schools, personal robots are expected to capture the attention of a new generation of computer scientists and engineers by “embedding learning in an interesting physical thing that moves around,” Balch says.

Home may become an easier and more pleasant place to live with personal robots that “take away the boring parts of life,” Christensen says.

As robots become more commonplace in people's lives, society must address the ethical questions.

“Researchers must involve many others -- such as philosophers and priests -- to contribute to the understanding of the relationship between humans and machines,” Christensen says. “If we just address these issues as computer scientists and engineers, we may come up with robots that look like what Hollywood creates.”