Crossroads - Winter 2016

XRDS • WINTER 2016 • VOL. 23 • NO. 2 even though that may not be the best way to solve the problem. You should really love the problem that you’re solving—not the specific tool—and look at the simplest way to solve that. Just keep on refining your thought process until you find out which ones are worth pursuing. Then you can build prototypes, simulations.

What technical challenges did you encounter when building things at Kiva?

Overall, what made Kiva successful was the fact that it worked. The biggest challenge we had at Kiva was taking existing technologies, subsystems, and combining them in just the right away to make a robust, reliable system. There is no single way to do it. It’s the art and the design of systems.

What are your views on how automation will impact the labor market?

I don’t defend automation from
claims of leading to job losses. I
freely admit that robots and auto-
mation replace human workers. If I
take a step back, and I look at eth-
ics and philosophy, what we should
be doing… we should shape our so-
ciety so that we don’t replace jobs
that people enjoy doing. What is the
purpose of life? Why do we live? It’s
not so that we can feed a big machin-
ery of consumerism. It’s so that we
can enjoy our lives. So we should be
striving to set up our society, our
rules, our incentives, so that people
do jobs that they enjoy doing. We
should strive to create the rules and
regulations that move our society to
where we want it to be. Currently,
do robots replace jobs? Yes they do.
Fortunately, they replace jobs that
for the most part are not very good
jobs. So that’s a good thing, right? If
somebody is sitting at a tollbooth,
handing out tickets to vehicles that
go through a toll road, to me that is
a job you want to automate. That’s
not a very good way for a person to
spend 30 years of their life. There’s a
second aspect here: Would a person
rather be doing that than being on
welfare? I think most people would
rather be working than being on
welfare. But then it’s a social issue.
When we get rid of these menial jobs
that are not rewarding, let’s make
to create things using technology.
I’m very good at math, physics, algo-
rithms, computer science, and ana-
lytical things. That’s the tool set; and I
like to create, so it’s only natural that I
use those tools to help me create.

Why did you choose to create autonomous systems for entertainment?

I just saw this as yet another way in which I could express my creative freedom. Verity Studios is just a way for us to have flying machines out there in the real world, doing sophisticated things. There are no killer apps for indoor flying machines yet, except for in entertainment. So this is a great way for us to get a head start on everyone, and actually do flying things indoors.

When did you first start building this stuff?

The first sophisticated autonomous thing I built was when I was an undergraduate student. We built a voice-controlled robotic arm as part of our undergraduate projects back in 1988-1989. However, my Ph.D. was in control systems. But, control systems, robotics, and autonomous systems are very tightly intertwined.

How did you graduate from building simple, low-cost machines to state-of-the art, precise, expensive ones?

On the contrary, the reason Kiva was successful as a company was because we figured out a way to make inexpensive robots. To have thousands of them in warehouses and to have them behave like they were expensive robots was the magic of system architecture, algorithms, and the components that we used. The robots were not expensive.

Can you elaborate on the differences between autonomous systems and robots?

The main distinction is that robots are subsystems of autonomous systems and they tend to imply motion.

Whereas a thermostat that regulates the temperature in your house is an autonomous system, there is no motion so people wouldn’t call it a robot. Cruise control in your car, people would call that an autonomous system,

but they probably won’t call it a robot simply because the car is not changing shape or doing manipulations.

How did you make that transition from working on simpler forms of autonomous systems to a more distributed network of robots?

I was working on distributed control when I was a professor at Cornell University. I ran a RoboCup project where we had mobile robots playing soccer. So there was a transition from things that you cannot do in a university environment, because you do not have the resources. But if you have a viable business model, then you can raise money, hire engineers, scientists, a sales force, and make it self sustaining. There are different challenges, but they share a lot of similarities.

Let’s talk about your creative process. How do you go from an idea to creating a blueprint for something you can build?

I’ll answer something different first. Coming up with ideas is easy. The first part is figuring out which idea is worth pursuing and then executing on it. Also, I just don’t want to discount the fact that it’s really hard to figure out what you should be doing. And once you’ve figured that out, once you assume that you want to solve the problem, then it really becomes about the problem. And this is where you have to be very honest with yourself and do something that engineers don’t like to do. Engineers like to come up with solutions. And they get enamored with those solutions