Architecture Journal Profile: Microsoft Robotics Group

While not widely considered in the enterprise except within certain verticals, such as manufacturing, robotics technology is today available to a broad software developer segment and is expected to be more and more present in the enterprise architectural landscape. With that in mind, we interviewed two thought leaders in the Microsoft Robotics Group about their careers and the future of robotics in the enterprise.

AJ: Who are you, and what do you do?

TT: I’m Tandy Trower. I’m the general manager of the Microsoft Robotics Group and founder of the Robotics initiative at Microsoft.

HFN: My name is Henrik Frystyk Nielsen. I am the group program manager in the Microsoft Robotics Group, and I’m very excited about bringing robotics into line-of-business (LOB) applications.

AJ: Robotics would seem to be a far remove from traditional LOB applications. Where do these domains meet?

TT: Let me answer that by describing the origins of the robotics initiative. The robotics community is quite diverse, from research on cutting-edge technologies (such as the Dynamic Source Routing protocol in the DARPA urban challenge) to very early segments of the commercial industries, as well as the existing industrial automation segment that’s looking for new marketplaces. The robotics community, through its various representatives, asked Microsoft for software assets that would address such challenges as maintainability and distributed concurrent applications.

When I presented this proposal first to Bill Gates and Craig Mundie, Craig pointed out that he had been incubating a new piece of technology to deal with the challenges of loosely coupled asynchronous distributed software. Henrik was a part of the team that created these. Although Henrik’s team did not have robotics in mind originally, the problem set that they were solving was very similar to the problem set the robotics community was facing.

These technologies, known today as CCR (Concurrency and Coordination Runtime) and DSS (Decentralized Software Services), form the foundation of the robotics toolkit. The core services of the robotics toolkit, to this day, have remained a general purpose programming application to deal with asynchronous applications. CCR is the piece that allows you to grapple easily with the challenges of writing an application on a multicore or multiprocessor system. DSS is that companion piece for scaling out across the network that allows you to have a consistent programming model locally and across the network and across heterogeneous types of devices.

As soon as we made the robotics toolkit available to the marketplace, customers outside of the robotics space started taking a look and kind of lifted up the hood and said, “That engine is good for my kind of applications, too. I have the same kind of problem.”

HFN: We recognized early on that more and more applications today are connections of loosely coupled components, and the challenge is about orchestration, about services. We wanted to bring a high-level application model to bear that inherently knows how to stitch together loosely coupled components. Bad things can happen—each component can fail at any point in time—but you want the application to survive, you don’t want it to just roll over and die. And at the same time we had to deal with and harness concurrency for better scale, better responsiveness, and applications that don’t fail.

Some customers recognized their own problem space based on the description of the basic components. For example, financial trading systems people said, “This is exactly about lots of data coming in, figuring out when to trade, and when not to, with independent agencies looking at the data with different viewpoints and decision making process flows”—a complex application that turns out to be similar to a robotics application where your information comes from the wheels, the lasers, and whatever else, and you have to decide, should I stop or go.

Other customers came from the robotics side. Robotics in fact is a great enabler. We had hobbyist roboticists come to us saying, “I’ve played with this toolkit over the weekend to build a robot, then I came into work and realized that I could apply the same matrices that I had done over the weekend to solve the problem in the enterprise.”

TT: Until now, we’ve bundled these assets into the robotics toolkit, but increasingly, customers outside the robotics space are saying, “These are very valuable assets, why do we have to buy a robotics toolkit to get them?” So, we will be addressing that by offering these components in a more palatable way to those customers; rather than having to buy them under the robotics cover, they’ll be able to get those assets directly.

AJ: What advice, not necessarily related to robotics, would you share with aspiring architects?

TT: Look for where things are going, keep an eye open to new opportunities. Certainly, at Microsoft there are leadership opportunities and opportunities for different kinds of leaders. Some leaders, like myself, are better in startup types of activities; others are better at maintaining stable organizations that are mature.

HFN: When you get to an expert level of understanding in a certain area, there’s a tendency to dive into that area and forget some of the other areas around you. You naturally build cultural walls between your area of knowledge and what you think other areas are doing. The experience of our team certainly demonstrates the benefit of breaking down those walls, being able to step back and recognize there’s a community, that there’s very great overlap we better talk with each other. It’s a huge advantage and I think it’s something that characterizes a great leader.

AJ: The architect role requires understanding technical trends to get the best business benefit. How do you stay up to date?

TT: The Web provides almost unlimited ways for connecting—blogs, forums, feeds, virtual libraries, wikis. Attending related conferences is another good way. A lot of the people in this industry are very heads down and focused. I think part of staying current is simply taking the time to look up and notice what else is going on. You might find some synergies there that you never even considered.

HFN: Just 10, 15 years ago, it was about getting information; now the challenge has become filtering the information. When there’s so much information that you can’t possibly consume it, you have to rely on new mechanisms. Figuring out how to get high-quality information about the things that actually solve problems is a critical task. Blogs, social networks, and newsfeeds of all kinds are a start.

AJ: Name the most important person you have ever met in this industry. What made him or her so important?

TT: There are so many. Of course, Bill Gates: He’s the reason I’ve been at Microsoft for as long as I have. Also, our more direct sponsor, Craig Mundie. Craig has been my manager a number of different times and is one of the forward-looking executives at Microsoft, having the ability to see where things are going, and then allowing me to participate, through what he was doing, and actually innovate in some of those areas.

HFN: I would add from my perspective Tim Berners-Lee, inventor of the Web, whose story shows how a single person effectively changed how the world operates today—from how we to work to how we talk.

AJ: If you could turn the clock back on your career, would you do anything differently?

HFN: Obviously, you think of many of your choices as discrete decisions, so it’s funny how very planned and deliberate your career appears in hindsight. I’m fortunate to have had many opportunities come my way. If you’re in a position where many opportunities arise, chances are good that an avenue will be open for you when you are ready for the challenge. But if you are in a situation where nothing really comes by, then you might be ready for a long time before an opportunity materializes.

TT: When I think through all of my life experiences and career experiences, I’ve always had a chance to kind of reinvent myself; it has been very satisfying to work in new areas. If anything, I would advise anyone to just be more aware of the fact that things never remain the same, that the industry evolves, so don’t to get too stuck on anything, because the landscape is going to change. Microsoft has had to work very hard to maintain its edge, and it only does that by continuing to refresh itself and to look to new things. So, stay fluid, and roll with the punches.

AJ: What does the future look like? What do you hope to accomplish in the next few years?

TT: Well, the robotics toolkit is part of a fairly ambitious goal to create a platform that would allow a new industry to emerge. Robotics has traditionally been in the industrial sector; a new business is now emerging from the personal side of robotics. Personal robotics has not arrived yet, but in many ways, it maps to the early PC industry of the late 1970s: When PCs first entered the marketplace, they kind of looked like toys. There wasn’t a clear understanding of what they were useful for.

Ultimately, I see robotics everywhere—like PCs, a very pervasive and ubiquitous technology. That’s going to take time, but I do believe that we will start to see that happen in three to five years.

HFN: I agree. Robotics has been on the edge for a long time, but it’s getting close to taking off in dramatic ways. It’s is going to drive how we interact with information and how we communicate, in social settings and industrial settings.

TT: One marvel of human design is that we are built from very simple processors that are massively connected together. The present computing paradigm is as if we’ve created a brain out of a single neuron. We’ve done wonderful things with it, but now we have the potential to define a technology that’s analogous to multiple neurons interacting together in a network effect. That’s reflected also in our social networks, ecological networks, or whatever you want to talk about that’s in some sense living.

Robotics represents the paradigm shift in the evolution of PC technology. The future is not going to be a single core processor sitting on your desk. You can now use technologies like CCR and DSS to develop a new generation of applications. It’s going to open up some significant horizons. Robotics is just the starting point.

This article was published in the Architecture Journal, a print and online publication produced by Microsoft. For more articles from this publication, please visit the Architecture Journal Web site.