The QS World University Rankings process is one of the big enterprises for rating almost 1,500 universities across the globe, based on a wide range of criteria such as faculty-to-student ratio, citations per faculty, and other things. McGill ranked first in Canada (31st world-wide) a few positions ahead of the University of Toronto (which is great and where I got my PhD). The very top spots world-wide went to MIT, Cambridge and Stanford.

All such ranking schemes are tricky, which the relative importance of various factors being quite subjective. There are a lot of great schools in the world and even schools ranked below fifty are still fantastic, like CMU, or the The London School of Economics and Political Science (LSE; 56th).

https://www.topuniversities.com/university-rankings/world-university-rankings/2023

This post provides an example of a short unsatisfactory (yet impressive) dialogue with ChatGPT. When I posted it in early Dec 2022, ChatGPT was still not well known and this was kind of interesting. In light of all the excitement since then, this is not quite so informative.

What is the computational complexity of global localization?

The computational complexity of global localization refers to the amount of computational resources (such as time or memory) that are required to perform
(... blah blah blah at least 5 paragraphs of very vague stuff --gd)

Overall, the computational complexity of global localization can be quite high,
(Note that this sentence above basically incorrect, as it is imprecise which is a variant of being incorrect in a mathematical incorrect --gd)
depending on the specific scenario in which it is being performed. In general, it is important ... blah blah blah

Wow, that was very wordy.

I apologize if my response was wordy or difficult to understand. As a language model, my primary goal is to provide accurate and comprehensive information on the topics that I am asked about. However, I understand that sometimes this can result in longer or more complex responses. If you have any specific questions or concerns, please don't hesitate to ask and I will do my best to provide a clearer explanation.

That is an amusingly wordy response to the accusation of wordiness!

Thank you for your feedback. As a language model, my primary goal is to provide accurate and comprehensive information on the topics that I am asked about. In this case, I was trying to provide a thorough explanation of the computational complexity of global localization, which can be a complex topic. However, I understand that sometimes this can result in longer or more complex responses that may not be ideal in all situations. If you have any specific questions or concerns, please don't hesitate to ask and I will do my best to provide a clearer explanation.

Well, thank good that at least you aren't repetitive!

As a language model, my primary goal is to provide accurate and comprehensive information on the topics that I am asked about. I do not have the ability to browse the internet or access any external information, so I rely solely on the knowledge and training that I have received from OpenAI. I strive to provide unique and informative responses to each question that I am asked, and I am not programmed to repeat myself or provide identical answers to different questions. If you have any specific questions or concerns, please don't hesitate to ask and I will do my best to provide a helpful and informative response.

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 based not only on ambitious new goals we want to achieve, but due to 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.

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 and 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 impact our conception of the world and our conceptualization of our role in 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 defined by 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.

With several of my colleagues from the Samsung AI Center, as well as one of my graduate students and McGill colleagues, I was able to visit Japan, attend the International Conference on Robots and Systems (IROS) and visit Kyoto and Tokyo.

IROS was good, and for many of the attendees was the first in-person conference since the pandemic began. There were a number of excellent keynotes that provided a retrospective view of how the field has evolved. Human robot interaction was a significant theme of the conference, and there was also the anticipated array of impressive Japanese mechatronics at the trade show (as well as in some of the talks).

The always-engaging Professor Junaed Sattar from U Minnesota give a talk about gesture based underwater robotics, and the paper was nominated for a best paper award.

Our team also got to visit the Samsung Research lab in Yokohama, which was extremely impressive, but confidential.

We were lucky enough to have a bit of spare time to visit the big Buddha at Kamakura, as well as the really fantastic national museum in Tokyo.

Our team with Tokyo skyline

Buddha at Kamakura

Mount Fuji from the train to Kyoto

I have a big telescope for star gazing, but have been hankering for something portable and with a big field of view. With the advent of image stabilized binoculars, it looks like it’s time to take the plunge.

The Canon 10x42L is my pick, as it combines a solid and pleasing 10x magnification with a big 42mm front lens and uses Canon's top-quality L-series optics. It's a very expensive pair of binoculars, but seems supreme. Notably, one should note that with binoculars and telescopes, the temptation of a large magnification or aperture size, in a lower-cost but lower quality package, should be resisted. For optical equipment, quality optics is really worth it, and a low-cost option is generally a road to frustration and disappointment.

At magnifications greater than 7, holding binoculars steady becomes a real challenge, especially over longer time periods which makes the image stabilization so attractive. People who have the 10x42L IS model often say they could not imagine going back.

To save a bit of money, the 10x32 version is still very good, but lacks the waterproofing and L-series optics. It has a smaller aperture, but still sports the very valuable image stabilization.




A much lower cost option would be the Celestron Cometron Giant 15x70 Binoculars (or the related and perhaps-identical "Skymaster"). Both are reportedly slightly inconsistent in performance and have been reported to occasionally arrive improperly collimated (which means the image is blurry unless you take steps to adjust it). This is still an interesting purchase, especially if you can find it on sale (they've been seen as low as $70 CAD/$55 USD in 2021 and if I could find them again at that price I'd grab a pair). At that magnification one probably wants to put them on a tripod for extended viewing (hence the attraction of expensive stabilization), but for short viewing sessions then just resting the scope
your elbows on a solid surface should be okay

The International Conference on Robotics and Automation (ICRA) is the premiere event for the robotics research community. The 2022 meeting in Philadelphia returned the meeting to a strong semblance of what it used to be before the pandemic, most recently in 2019 (Montreal). In between the conference took place, but generally with a greatly-reduced in-person component, or a very atypical geographical distribution of attendees. The attendance in 2022 was as big or bigger that it had ever been, and the researchers and companies in attendance were greatly invigorated by the opportunity to see each other in person again, and to engage in the professional networking that is so important in building a community. In general, my impression is that smaller meetings and sub-communities have done pretty well in "virtual mode" during the pandemic, but building connections is a much larger community, especially between people who don't already know each other was difficult.

Robotics has continued to thrive all through the pandemic, and the quality of the research papers being presented was noticeably strong. For somebody like me who has been "in the game" a long time, it's exciting to see how much the quality of the work being done has improved, in terms of core science, engineering, testing and pragmatic impact. In the context of pragmatic impact, the trade show was also very impressive. The kind of companies on site, and the sophistication of the robotic systems and technology being show was head was wonderful. My own team from Samsung had a hiring booth and we got applications and expressions of interest from a students and professions of the highest calibre.

...