08 July

Robotics as a paradigm for discovery: Updated notes for the Berlin Robotics Summit

What is the role and destiny of science in the mid 21st century? What is the evolving relationship between AI, robotics, computer science, and traditional scientific inquiry? The biggest difference is that robotics (like AI and Computer Science) deals with developing new systems that build on one another, and while often being embedded in the real world, are largely preoccupied with understand non-physical interactions. In this article I see AI and robotics are deeply intertwined disciplines , and terms can be used roughly interchangeably.

Looking back to Ancient Greece, Physics (or “Natural Science” as it was more generally framed) played a critical unifying role in shaping humanity’s understanding of the natural world and our role within it. This unified vision was significant in that it allowed for an organized and coherent development of new ideas and source of knowledge (and power) that the Western world went on to develop over the subsequent two millennia; a process of intellectual growth akin to, and related to, developments elsewhere in the world. It has served as the intellectual framework that related mathematics, the physical world and our conception of how it works and how to control it. As Natural Science became known as “physics” (which was essentially synonymous with “all science”), and which eventually bifurcated into the allied specialties that make up the various natural sciences (such as Chemistry and Biology) it allowed scientists to determine what common bodies of basic knowledge constituted their domains, and to develop not only theories and models, but also shared curricula, academic programs and even political agenda.

Our field is now facing the same challenges in the domain of “synthetic science”: the science of the artificial, virtual and man-made systems that are already of enormous influence and importance. The constructs of synthetic science (such as a major computer operating system) are already as complex as anything that mankind has ever built, both in purely intellectual terms as well as in terms of actual artifacts. We need to not only develop tools and methodologies, but also to identify and formalize basic questions, and to circumscribe coherent new domains of discourse. Thus, as synthetic science progresses, new challenges are developing. These challenges re based not only on ambitious new goals we want to achieve, but also the complexity of the objects and ideas under consideration. Much of this new domain of science and engineering can be described by one broad term: robotics (although artificial intelligence – AI --- has an overlapping domain of inquiry, and in the medium term addresses roughly the same questions).

Robotics in its broadest form can be defined as the discipline concerned with both the development and modeling of systems that (1) make measurements of the real world, (2) perform computations, and then (3) act upon the real world in some substantial way. By this definition, more and more of the objects in our everyday world are becoming robots, and this is happening rapidly. This includes, of course, cell phones, cars, security systems, and many of the appliances in our homes. The microwave oven in my own home, for example, measures the weight and humidity of food we put into it, computes the appropriate cooking time and power levels needed, and then acts upon the food to cook it. As almost every object within our lives becomes computationally enabled, myriad new challenges are starting to emerge. As many devices start to become independently mobile, or interact with other devices that are mobile, these inherent challenges will increase substantially. As our culture is subsumed by robotics technologies, do we not need an all-embracing domain for this huge new body of challenges?

The implications of defining robotics as a broad umbrella are twofold: one pragmatic and one conceptual. The conceptual implications relate to the development and organization knowledge, the construction of pedagogical systems and programs of instruction, and the development of formal mathematical frameworks for very complex artificial or emergent systems. The pragmatic implications relate to the fight for funding, recognition of programs within our universities, and the ability to efficiently carry out our research.

Physics (or natural science) has been defined as the understanding of the “laws and phenomena of the natural world,” while traditional engineering deals with the application of that understanding to the creation of new artifacts. Our challenge in robotics is also to understand and predict the operative laws in our discipline, but they are not exclusively the laws of the “natural world”, and in fact we have the option to generate new laws (for example network protocols that govern information flow or connectivity). Thus, robotics is profoundly theoretical as well as distinctly experimental.

Moreover, a critical part of the robotics research enterprise is to build, measure and eventually control the artifacts we are envisioning. These steps are not always sequential: with networked systems, for example, we often observe unintended phenomena that must be understood after a system has been designed, built and is already under control.

Not only will (does) robotics and AI impact our conception of the world and our conceptualization of our role within it, robotics also has the potential to impact our very sense of identity. It is a domain that has already impacted notions of how people function and how biological organisms evolve. As such, robotics is reshaping not only our lives and our society in pragmatic terms, but also how we see the world and ourselves within. Is this not the same kind of conceptual reformulation that led to the Renaissance?

What is required is a unifying science of what will govern a critically important new world view. If robotics and related technologies have the impact we expect, and which in fact seems inevitable, then there can be no doubt it will impact our conception of science, engineering and society. One needs only to reflect on how notions of computing, computers and algorithms have shaped most areas of human though over the last 50 years, where computational ideas have fundamentally changed thinking in areas as diverse as biology, banking, dating, sculpture, communications and criminology.

In addition, is it clear that many important ethical and social issues are looming. They need to be addressed in a context that is technically broad and mature.

A topic of current discussion and debate both at the workshop and in society at large is the notion of “The Singularity,” as considerd by von Neumann, Vernor Vinge and Ray Kurzweil. While the singularity itself is a topic of substantial controversy and some doubt, the accelerating pace of technological change that is used to substantiate this notion is broadly agreed upon. This accelerating rate of change increases the need, and the urgency, of recognizing the role of robotics today and bringing the disparate ideas and disciplines involved into a coherent and collaborative framework.

Robotics is the branch of human endeavor that integrates both engineering and science, and cannot be pegged well in either alone. By subdividing the field into 2 different academic faculties (Science and Engineering) or disparate disciplines (Computer Science, Mechanical Engineering and Electrical Engineering), the additional potential for fruitful interaction is decreased precisely in a subject where this interaction is critical. In addition, it becomes more difficult to recognize a common body of prerequisites, knowledge, and tools that the students and practitioners would best be equipped with. In short, the divide between Science and Engineering is not appropriate to a domain of discourse defined by intellectual constructs that are created by human hands. Robotics, perhaps more than any other area of inquiry, falls on both sides of this divide and thus progress is directly impeded by the partition between traditional engineering and science.

Robotics has a fundamentally different (and broader) mandate from many classical areas of computer science like complexity theory, compilers, quantum information theory.

Much of classical science is reductionistic, but even the scientific part of robotics are not.

Artificial Intelligence, as a research domain without robotics, becomes increasingly arcane and irrelevant. Likewise Computer Vision without robotics would have to ignore fundamental issues of great value and importance.

Traditional academic disciplines like Computer Science, Mechanical Engineering, and Electrical Engineering are likely to be preoccupied primarily with embedded systems, smart machines, and self-diagnostics systems. Systems which are, in a deep sense, robotic systems. Moreover, systems which fundamentally and by their very nature cross the barriers between these narrow disciplines.

No other containment relationship between academic disciplines is as consistent as using robotics to refer to the high-level aggregate. Robotics cannot be a little niche within Mechanical Engineering or Computer Science, it just does not fit such narrow confines. On the other hand, while no strict hierarchical ordering of academic disciplines is perfect, making Robotics an umbrella discipline for several other sub-areas is probably very natural, and will become more so as the science and technologies of the discipline evolve.

We need to promote this coherent world view in education and government. Such a unified framing of the discipline is useful in the quest funding, student development, the consideration of ethical issues and other integrative-level issues.

This “speaker’s corner” discussion was based on a position statement from the author meant to generate some debate. It was followed by an open lightly-moderated and rather energetic discussion from the participants at the workshop. It related to the role robotics can play as a unifying banner for several areas of inquiry. Overall, the discussion seemed productive and several ideas were generated. These notes capture the main thrust of the position statement in prose form, and several of the implications and prods to discussion in abbreviated form. Due to the need to moderate and participate in the discussion, only limited written notes were captures from the open discussion although some of those thoughts have been used in the reformulation of the introductory text.

The discussion revolved around the potentially primal role that robotics seems destined to play in the intellectual lives of people in the next century. This was framed in the context of the history of science.

By Gregory Dudek at | Leave a comment |    
There are no comments.
Post your own response

Each comment is manually screened for the presence of appropriate and substantive content, due to a constant onslaught of comment-spam. This means there may be a delay before your comment appears.

(Some kind of name is required, will be visible)

Required, whatever you enter will be visible to other users.

(Optional, used for "mailto" link)

Your email address is not required, but if you insert it it will be displayed so people can contact you.

Answer this question correctly to demonstrate that you are not a dumb spambot.

The title for your comment.

Your comment goes here. All relevant comments are welcome, except for those that simply promote an irrelevant product or else are used to fraudulently inflate the link count to an irrelevant web page. They appear after moderation. Don't forget to also fill in the captcha below or your text will be rejected automatically!

You must answer this question to prove you are human
Which one has ice cream (pick a number 1-5): 1:DairyQueen, 2:KFC, 3:FBI, 4:NFL, 5:NBC-TV?

Answer this question correctly to demonstrate that you are not a dumb spambot.