Siebel Scholar Jingjin Yu (UIUC, CS ’08) is working to solve that problem by developing algorithms that enable swarms of robots to work better together. Robots aren’t exactly taking over—not yet, anyway—but their numbers are growing. In the next two years, the population of industrial robots will double to nearly 3 million globally. These busy machines move parts around factory floors and select items in warehouses. But more robots in the same space means a higher likelihood of collisions and even traffic jams.

An assistant professor in the Department of Computer Science at Rutgers, the State University of New Jersey, Yu focuses on solving the computational problems that will prevent robot-to-robot collisions and increase their efficiency. His research allows any machine that needs to plan a route through a crowded space to make better decisions and accomplish tasks faster, with less missteps.

Yu’s work reaches beyond the factory floor and could change the way robots patrol borders to perform surveillance or explore new territories to collect data. His work has applications in any industry where driverless vehicles like drone aircraft or autonomous automobiles need to navigate airways or roads. And for more traditional companies that use fleets of standard human-driven cars or trucks to deliver goods and services, Yu’s calculations can determine the best way to coordinate the fleet and avoid traffic jams.

Although Yu’s expertise would be an attractive asset to an array of companies, the academic environment is where he feels he has the most freedom to innovate.

Yu spoke with the Siebel Scholars program about his journey from his hometown in Anhui, China to researching a complex problem and teaching at an elite university.

Q: Can you give some examples of places where a large number of robots need to coordinate to move in crowded spaces?

At Amazon, they have a bunch of not quite small but not quite big robots, maybe 2 -3 feet long and 2 -3 feet wide, working in their order fulfillment centers. For you and me to get orders from Amazon, someone has to go pick the object off the shelf. For people to make the trip to the shelves, it takes a lot of time, so they use the robots to bring the shelves to the pickers’ stations and it’s a lot faster. The goal is to pack as many robots into a crowded space and to get to the shelves as quickly as possible.

Q: Do robots present a unique challenge?

A. Definitely. Multi-robot systems create a very interesting computational challenge and the problem appears in our everyday life all the time. As an analogy, think of a shopping mall or a traffic hub. When there are few people around, they can all easily get where they want to go as fast as they could walk. However, during weekends at shopping malls or rush hours at traffic hubs, when the density of the crowd increases, this is no longer the case. The same applies to road traffic, for example, at intersections with four-way stop signs. Here, when people are replaced with cars, the problem becomes even harder: while it is generally OK for people to bump into each other, it is probably a bad idea for cars to do so. Robots are more like cars than people in this case.

Referring back to the Amazon fulfillment center example, it would be beneficial to have more robots moving things around if they can also move around quickly. So the unique and challenging computational problem we hope to address is: how to maintain the navigation efficiency when the density of robots is very high?

Q: Does your work also touch on the way humans can get where they need to go in the shortest amount of time?

Yes, imagine a problem with how a tourist will tour a city. As a tourist, you go to a new city, maybe for a business trip. You don’t know the city too well. Suddenly you have a few hours of time so how do you plan an interesting trip through the city? There are famous places you can go. You can go to one place and stay for some time and get some satisfaction. Or you can go to several places and stay for shorter periods of time and that’s satisfying too. But how do you determine the best way to use your time?

This gives us a problem that looks like this—you are at a hotel, you have 20 points of interest and you cannot hope to go to all of them. You want to find a good subset of places you can go and plan a trip that maybe includes taking the bus to Point A, staying half an hour and taking another bus to Point B and so on. This can be thought of as a robotics problem and is similar to the situation if you a quad copter or a drone and you wanted to use it to check how traffic looks at major intersections.

Like the tourist, the drone has to go to various places, in this case intersections, and stay for some time to collect information. So it’s the same problem. The drone is making observations and you want to maximize information collected by the drone, same as the tourist would want to maximize pleasure from the trip.

Q: It sounds like there are many commercial applications of your work. Have you contemplated going to work in the private sector?

In an academic setting, I get freedom to do things that I wouldn’t get to do in a company. I have the opportunity to do industrial things but the way I can make a more valuable contribution is to do research on the math and advance general knowledge that can be transferred to many different applications.

Q: You’ve been at Rutgers nearly two years and you teach both undergraduate and graduate students. Can you tell me what that’s like?

Teaching is quite interesting and I get to learn a subject a second time, often more deeply, at the same time. In smaller graduate classes, where I have the chance to know the students, I can challenge their individual ability to the extreme and pose difficult questions. Someone will hopefully come up with a good answer and if not, we’ll get to think about it.

With undergrads, the challenge can be finding out how to give a good course coverage to a very large group, where each student has a unique learning history. Last semester I taught Discreet structures, a mathematics class, with 140 students.

Q: In addition to teaching, you work on several research projects at once. Can you tell me how that works?

Research is interesting but it is an exploration into the unknown. I work on the computational side so I usually juggle among a number of such problems. These types of problems can be rather hard to know whether you can actually find good algorithms for solving them nicely. It’s good to have a few things to work on where I know that I’ll be making progress on some. It’s also refreshing to switch between topics once every few hours or days.

Q: Did you know from a young age that you wanted to study robotics and computational problems?

No. I took an interesting path. I was trained as a chemist as an undergrad at the University of Science and Technology of China. I came to the US, first to University of Chicago, to do my first Ph.D. study in Chemistry. I enjoyed playing with arranging molecules but you have to set up experiments and wait there, sometimes for a few days. At times, you have little control over what is going to happen. That aspect of the experience was a bit frustrating.

After that, I decided to get some industry experience. I worked at AT&T for a few years. Then I worked at a hedge fund for a year where I did front desk support, providing technical expertise to support the traders. I could probably have stayed and become a trader at some point but it’s not what I wanted. I realized I still liked academia.

I came to realize that computer science was a subject I had a true passion for. Even though I did not have a formal computer science training, I enrolled at University of Illinois, and in first four courses, I received two A’s and A+’s.

Q: You were a Siebel Scholar during that year, when you started to pursue your Ph.D. in Computer Science. What was that like?

It’s quite prestigious. I remember I went to a meeting that year in Berkeley and I enjoyed talking to all the other Siebel Scholars. It was amazing to see that Thomas Siebel himself cares about so many different areas of research and all kinds of aspects of humanity. Quite a few of the Siebel Scholars are from the business side and since I’m from the tech side, it was a great experience to hear different ways of thinking.

Thomas Siebel was an Illinois grad too. When I was still in graduate school, I met many other scholars through local meetings. It was a great experience.