|September 12, 2005|
Research News Editor Eric Smalley carried out an email conversation with
Georgia Institute of Technology professor Ronald
C. Arkin in August of 2005 that covered the economics of labor, making
robots as reliable as cars, getting robots to trust people, biorobotics,
finding the boundaries of intimate relationships with robots, how much to
let robots manipulate people, giving robots a conscience, robots as humane
soldiers and The Butlerian Jihad.
Arkin is Regents' Professor in the College of Computing at the Georgia Institute of Technology and the Director of the school's Mobile Robot Laboratory.
Arkin's research interests include behavior-based reactive control and action-oriented perception for mobile robots and unmanned aerial vehicles, hybrid deliberative/reactive software architectures, robot survivability, multiagent robotic systems, biorobotics, human-robot interaction, and learning in autonomous systems. He's written more than 120 technical papers, a textbook entitled Behavior-Based Robotics, and co-edited the book Robot Colonies.
He is an IEEE fellow, co-chair of the IEEE RAS Technical Committee on Robot Ethics, associate editor of the journals IEEE Intelligent Systems and the Journal of Environmentally Conscious Manufacturing, and Series Editor for the MIT Press book series Intelligent Robotics and Autonomous Agents. He served on the National Science Foundation's Robotics Council from 2001-2002.
Arkin has a bachelor of science degree from the University of Michigan, a master of science degree from Stevens Institute of Technology, and a computer's science doctorate from the University of Massachusetts, Amherst.
TRN: What got you interested in science and technology?
Arkin: Good question. I always have been from as early as I can remember. Maybe it was the International Geophysical Year (IGY), Mr. Wizard (an early TV Science Guru for kids), or just my own innate curiosity. Whatever it was it has been a passion since my youth.
TRN: What are the important or significant trends you see in science and technology research?
Arkin: Pretty vague question there, and of course since S&T is such a broad area, I'll probably focus on what I know best.
Certainly computational power has advanced so dramatically in recent times that things that were unapproachable a few decades ago are now commonplace. With the advent of supercomputing (there's one right next door to me while I sit here in Tokyo) fundamentally new approaches to problems can be taken.
TRN: Tell me more about what ever-increasing computational power is likely to enable and how supercomputing will enable fundamentally new approaches to problems.
Arkin: Some of my colleagues feel more strongly than I do on this subject, e.g., interpreting Hans Moravec; it would seem that it's simply a matter of time that advances in computational speed will enable human-level intelligence in machines.
My tack is somewhat different... I believe that now with better tools, we can explore our understanding, development and implementation of intelligence in novel ways, perhaps, for example, by creating highly efficient distributed and parallel machines that can enable us to better understand and recreate intelligence. The end result will be new paradigms.
What specifically they will be remains to be seen, but just as better ships enabled man to voyage further on his expeditions, even into space, so will better computational engines facilitate new discoveries of this sort.
TRN: Tell me about trends in robotics. What are the pluses and minuses of the technologies as they exist today? What do you see as the most urgent needs in robotics research and development?
Arkin: The field seems to go in spurts - new paradigms emerge, then for several years they're refined, then the next significant breakthrough occurs.
Either fortunately or unfortunately depending on where you are sitting, there are still plenty more things waiting to be discovered - fundamental knowledge regarding the machinations of intelligence are still lacking.
The good news is that we are constantly gaining more knowledge and models of biological intelligence, which for me and many others provides fodder for creating truly intelligent machines. A deeper understanding of process models of human-level intelligence would be quite helpful, as well as a deeper understanding of how humans and robots can and should relate to each other is another pressing need.
TRN: What is the current understanding of human-robot interaction and what are some of the more interesting ideas in this area?
Arkin: It's quite limited at this time, as real world robots are just beginning to make their appearance in everyday life as they move out of the laboratory.
Formal studies, not unlike those found in the human-computer interaction (HCI) and the man-machine interface (MMI) communities need to be conducted, and many of us at this time are pioneering the experimental methods by which these studies should be performed. As robots can appear to be intelligent cognitive actors to people, it changes the equation somewhat from more traditional computation or automation.
Some of the interesting questions include:
· What are the appropriate relationships between humans and robots?
· Are there significant cultural variations in the acceptance of this technology?
· Can these machines be used to manipulate human thinking and behavior?
· What should these agents look like and what is appropriate behavior for them?
· What are the most appropriate modalities for interaction between robot and human?
TRN: Much has been made lately of the emergence of a retail market for robots, which so far has been well-automated vacuum cleaners and humanoid robots that can be described as very expensive toys.
How close are we technologically to everyday encounters with robots in the home, office, shopping mall and city street? What are the missing pieces that will get us there?
Arkin: Simple economics. People are still cheap, flexible, and versatile when compared to robots.
In many respects the basic science is already present for a much larger presence of robotic technology in the everyday world. Most commercial robot ventures still currently fit the old paradigm of the 3Ds: Dull, Dangerous or Dirty tasks, which people do not want to do. The military is exploring robotics research extensively, but in the battlefield and not in the service sector.
Perhaps the most conspicuously missing component for acceptance is reliability. People want 24/7 operations - these machines must be as reliable as our automobiles. We are not there yet for almost all robotic applications.
TRN: What needs to happen to get there?
Arkin: I guess the same sort of reliability and systems engineering that has been applied to traditional automation such as cars, airplanes, escalators and the like, and major companies that are capable of providing that level of engineering and experience need to get involved.
TRN: It seems likely that in the next 10 to 20 years or more, most robots in the field will be based on vehicles or have bodies that do not resemble humans, and their high-level capabilities will be limited and task specific. What will be impressive about these machines?
Arkin: Bit of an oddly put question. Robotics should be designed for the tasks they need to accomplish and the environment in which they reside in. In most cases, this does not require a humanoid form. A trash compactor could look like a person, but why should it?
Intelligence and mobility is not solely the province of mankind - we share it with many other biological systems that do not have a human form. There is no reason to believe that robots should be largely human as well. If they accomplish the task they were designed for in a reliable and effective manner no matter what their shape, these are significant achievements. What makes them robotic is their ability to sense and respond intelligently to the environment.
Humanoid robots are best suited for human-robot interaction, as it seems more likely that people will interact with a human-shaped robot than one that looks like a trashcan.
TRN: In many ways, the technology for robotic motor vehicles is well-developed. What's needed to make them safe enough to share the road with people?
Arkin: Safety and reliability engineering is a field unto itself. Every time I go to the Atlanta airport I get into a train without a driver to take me to the terminal, yet I think nothing of it, as do the millions of other people who have shared the same ride.
Acceptance of new technology takes time and the development of trust in users. Issues in legal responsibility for the presence of autonomous highway vehicles mixed with human drivers will also need to be addressed at some point.
In many ways I suspect that many drivers out there on the freeways are far more dangerous than a predictable autonomous vehicle might be. The biggest difficulty is that if robotic vehicles are mixed in with regular traffic, either inadvertent careless or malicious behavior on the part of human drivers could create situations for the autonomous system which are quite unsafe.
These exceptional situations are the ones most difficult to deal with. Also coping with all types of lighting and weather conditions poses significant difficulties as well.
TRN: The concern about human safety around robots is easy to understand. But what about the other way around? I picture the fate of a humanoid robot that attempts a rush-hour subway ride in Tokyo or New York. How do you give robots the ability to protect themselves or at least make them durable without making them dangerous?
Arkin: Good question and indeed one of my students is working in this area. It speaks to the issue of trust. Most people ask how do we make people trust robots, but the other question is also valid, how do we get robots to trust people?
People can be deceitful, malicious, mischievous, and otherwise malevolent either intentionally or unintentionally. By looking at models of how humans create and maintain trust, we can begin to implement computational models in our robots so they can learn not to take commands that would put them in danger for example.
TRN: Much of robotics research is aimed at developing and coordinating low-level behaviors, like giving robots the equivalent of reflexes. Can you give me a sense of the challenges involved and some of the milestones to watch for?
Arkin: This has been an area that has been richly mined over the last two decades. Many different approaches have been developed (I can plug my Behavior-Based Robotics book here) that provide effective mechanisms for combining low-level reflexes and behaviors into intelligent emergent behavior.
Some of the outstanding issues deal with hybrid architectures - just how is intelligent planning and reasoning integrated with low-level behavior. Many approaches are still being studied, including some new methods being developed during my stay here in Tokyo at the Sony Intelligence Dynamics Laboratory. I am optimistic that significant advances are not far away in this regard as we move closer to the current pinnacle of human-level intelligence.
TRN: Tell me about biorobotics.
Arkin: What do you want to know? That's a gigantic area.
TRN: Biorobotics is aimed at applying knowledge about biological systems to the development of robots. One obvious area is locomotion, given that humans, snakes and insects are better at moving in certain environments than wheeled or treaded vehicles. What are some other aspects of biology that are helping us make better robots?
Arkin: Virtually every aspect of robotics is touched by biology. This ranges from psychology to ethology to neuroscience to name just a few that have influenced my own research. There are whole conferences on this topic, and it's a pervasive influence. A more appropriate question might be what aspects of biology have not had an influence - even seemingly esoteric fields such as ontogeny, immunology, and endocrinology have had their impact in the robotics research community.
TRN: How do the technologies you're working on relate to business, culture, and social life?
Arkin: I am currently very active in several research communities within robotics: multi-robot systems, human-robot interaction, and robot ethics.
In many ways these areas are all inter-related: my early work with teams of robots, led me to consider humans as members of these teams. How can we effectively engage humans with robots and vice versa? This is crucially important for their acceptance in the marketplace and everyday life. Next year the first Annual Conference on Human-Robot Interaction will be held in Salt Lake City, and other new fora such as this will help provide answers to your question.
Ethical considerations are a more recent concern on my part, and are an outgrowth of my interest in human-robot interaction. Having participated recently in several workshops and symposia on the topic, and being involved with the IEEE Robotics and Automation Society's Robot Ethics Committees, I have looked back on my life's work thus far with considerable introspection in this regard.
Much of the technology we are creating goes unchallenged and is entirely curiosity driven. While this is the classic research model, we must not let curiosity get the better of us, and we should take the time to consider the consequences of the artifacts and applications we are creating, particularly in terms of human relations and dignity, religion, professional conduct, warfare, etc. I recently offered a course at Georgia Tech on the subject entitled Robots and Society that shared these issues with our undergraduates. I hope to write a textbook on the subject in the near future. So in maybe a year my answer to your question will be even more definitive.
TRN: What are the important social questions related to today's cutting-edge technologies?
Arkin: What are the boundaries, if any, between human-robot relationships? What is the role of lethality in the deployment of autonomous systems by the military?
TRN: So what are the boundaries between human-robot relationships?
Arkin: I tend not to be prescriptive about these boundaries, that's a question of morality. I am interested in the ethical issues surrounding these questions though, which will lead to the formulation perhaps of a moral code some day.
A few of the human-robot interface questions that concern me include:
· How intimate should a relationship be with an intelligent artifact?
· Should a robot be able to mislead or manipulate human intelligence?
· What, if any, level of force is acceptable in physically managing humans by robotic systems?
· What do major religions think about the prospect of intelligent humanoids? (The Vatican and Judaism to date have had related commentary on the subject).
These are all ethical questions, and depending upon your social convention, religious beliefs, or moral bias, every individual can articulate their opinion. My concern now as a scientist that is concerned with the ethical outcome of his own research is to get the debate going and begin to explore what the appropriate use of this technology from a variety of ethical stances (relativistic to deontological).
TRN: And what is the role of lethality in the deployment of autonomous systems by the military?
Arkin: [This is a question] I hope to personally pursue more deeply from an ethical perspective, especially as a large portion of my research has been funded by the Department of Defense.
The real issue is whether the robot is simply a tool of the warfighter, in which case it would seem to answer to the morality of conventional weapons, or whether instead it is an active autonomous agent tasked with making life or death decisions in the battlefield without human intervention. To what standards should a system of this sort be held to, and where does accountability lie?
In addition is it possible to endow these systems with a “conscience” that would reflect the rules of engagement, battlefield protocols such as the Geneva Convention, and other doctrinal aspects that would perhaps make them more “humane” soldiers than humans? I find this prospect intriguing.
TRN: In terms of technology and anything affected by technology, what will be different about our world in five years? In 10? In 50? What will have surprised us in 10 years, in 50?
Arkin: I've never been a prognosticator or futurist and am not about to start now. As a scientist I prefer to have a firm basis for my projections, and going 10-50 years out in this day and age is beyond my capabilities. In that sense I'm not as bold, or foolish, as some of my colleagues.
What I can say is that robots will continue to become more commonplace, and they will be accepted more readily, starting to vanish into the background noise of technology that many devices already occupy, such as aircraft, televisions, air conditioners, etc. Truly intelligent robots will stand out, but when that will happen is not clear. But not in 5 years.
TRN: What do you imagine you or your successor will be working on in 10 years? In 20 years?
Arkin: Gosh, if I knew that I'd be working on it now.
TRN: What's the most important piece of advice you can give to a child who shows interest in science and technology?
Arkin: The best advice is to the parents: encourage it and let it grow. Directed curiosity is a wonderful thing in humanity: discovery and learning and acquisition of knowledge can provide a lifetime of satisfaction. And you can make a decent living as well.
TRN: What's the most important piece of advice you can give to a college student who shows interest in science and technology?
Arkin: Pay attention to basics. Defer gratification until you have mastered the fundamentals of mathematics, physics, and the other disciplines. Also pay attention to interdisciplinary studies - there's much to be learned by being a generalist. Also watch and learn from your more senior counterparts and find good role models.
TRN: What are your thoughts on the state of conventional wisdom on science and technology?
Arkin: Not sure I full understand the question. I believe in the scientific method as the basis for conducting meaningful investigations.
I was trained as a chemist before I became a computer scientist, and my background in formal scientific methods, experimental design, hypothesis testing, and analysis I believe has been of great value and a decided advantage to me as a computer scientist and roboticist. It unfortunately is generally woefully lacking as part of a traditional computer science curriculum.
TRN: What could be done to improve the pursuit of science and technology research in terms of business trends, politics, and/or social trends?
Arkin: Funding of course enables discoveries, but does not guarantee they will occur. Lack of funding can almost certainly guarantee that discoveries will not be made.
A renewed emphasis on long-term research as opposed to the short-term and heavily applied research funding styles that are currently en vogue would help greatly in that regard and ensure that we can continue to produce the intellectual capital of the future for industry, government and academia.
TRN: What books that have some connection to science or technology have impressed you in some way, and why?
Arkin: The Dune Prequels co-written by Frank Herbert's son are an interesting read [The Butlerian Jihad, The Machine Crusade, The Battle of Corrin], especially those aspects surrounding the so-called Butlerian Jihad, where mankind is struggling to free itself from the tyranny of a universe of robot domination.
It's an interesting perspective on how the designers of intelligent systems can create the antithesis of the Utopian Society that many envision.
TRN: What other readings do you recommend that would bring about more interest and/or a better understanding of science and technology?
Arkin: Wilson and Holldobler's “The Ants” shows how very simple intelligence can lead to massive accomplishments even when the coordination between entities is quite simple. And for a nice high-altitude view of current robotics research Menzel and D'Aluisio's book Robo Sapiens is more than adequate.
TRN: Is there a particular image (or images) related to science or technology that you find particularly compelling or instructive? Why do you like it; why do you find it compelling or instructive?
Arkin: There's one from an old National Geographic that I included in my book that was taken by an undersea scuba diver in Australia showing the interaction between a flock of diving birds and a school of anchovies. The natural geometries help to validate some of the theoretical models we consider.
TRN: What are your interests outside of work, and how do they inform how you understand and think about of science and technology?
Arkin: I enjoy reading the scriptures on a daily basis, and am perfectly comfortable with the relationship between Christianity and science. I enjoy travel immensely as well, although I probably do far too much of it for my own good, as I'm now in Tokyo until October and then head to Toulouse France for 10 months. Seeing and interacting with other cultures helps me both personally and provides a basis for a broader understanding of robots' place within humanity.
TRN: Is there anything else you would like to say?
Arkin: Thanks for the opportunity.
TRN needs your help. Please click here for details.
Research News Roundup
Research Watch blog
View from the High Ground Q&A
How It Works
New: TRN's Jobs Center
News | Blog | Books
Information on RFID, VoIP, VPN, ZigBee and more
Buy an ad link
Ad links: Clear History
Buy an ad link
© Copyright Technology Research News, LLC 2000-2006. All rights reserved.