Three small quadrotor drones execute a series of synchronized flips in the middle of a Gantcher Center tennis court. This performance, captured in a minute-long Youtube video uploaded last summer, showcases the work of Tufts’ Signal Processing and RoboTic Networks (SPARTN) lab. Since its establishment in 2012, the lab has successfully brought together theories and applications as it incorporates students from all levels.
Started by Assistant Professor of Electrical and Computer Engineering Usman Khan, the lab contributes to Tufts’ signal processing and robotics research. It currently focuses on “multi-agent networked systems,” and uses the Parrot AR.Drone platform as a test-bed, according to its website.
“The research that I do is more theoretical, so my Ph.D. students work on very abstract problems,” Khan said. “The concepts are mathematical and statistical, so it’s hard to explain the research to someone who is not experienced in those areas. With SPARTN I wanted to create a lab that ‘puts [an accessible] face’ on the abstract research. That is the most important thing here.”
Michael Tran, who is currently a graduate student at Tufts, joined the SPARTN Lab in its first year as a rising junior. He explained the goal of using the drones.
“Professor Khan’s area of expertise is wireless sensor networks and distributing systems, so his end goal was using these drones as a platform test for algorithms and other ideas,” Tran said. “The idea is to use these drones as a testbed for wireless sensor networks.”
The lab is also used to expose undergraduates to research, according to Khan.
“The lab gives me an opportunity to interact with and integrate undergraduates students in the research,” Khan said. “The way we work the abstract theory is developed by the Ph.D. students or master’s students who have seen this research for a while, and the undergraduate students do these smaller projects which show some the ideas we have developed in our research.”
Tran believes that working with SPARTN Lab gave him experiences he did not find in the classroom.
“Compared to a lot of the stuff I did in the classroom, SPARTN Lab was more hands-on,” Tran said. “The project that I was working on was with these quadcopters that Professor Khan ordered. Our job was to actually program them. In class I had learned to program stuff, but applying it to an actual robot was something you don’t really get in the classroom.”
Undergraduates are given as much autonomy as they want while working in the lab, according to Khan.
“The autonomy for undergraduate students ranges from zero to 100 percent,” Khan said.“We have done some projects where it has been completely autonomous.”
Often, Khan will provide the equipment and allow undergrads to devise their own projects.
“I have said, ‘This is what I can provide — quadcopters, computers — and I can buy some equipment, like microcontrollers or wireless communication chips or GPS devices,'” Khan said. “Then they come up with their own project, so a lot of the projects we did initially, the undergraduate students came up with. I said, ‘This is the equipment that we have, this is the direction that we want to go, now what can you do in this domain?’”
The Ph.D. students are also responsible for setting the framework and goals for the projects in the SPARTN Lab, according to Ph.D. student Chenguang Xi.
“The most important thing is to let the undergraduate students know how everything works, and then the details of implementing the theory, they will figure out for themselves,” Xi said. “We need to let them know what the aim of the project is, to set what the big picture is.
Xi stressed that it was important that undergraduates understand the abstract concepts that the projects involve.
“We focus more on what the theory is behind the application, because it is very important to understand those things first,” Xi said. “Once they understand those things, they may only need to be told the final goal. That could be only one equation, and they are programming to implement that equation. Then it’s not our job, it’s their work.”
One of the lab’s current projects involves the use of quadcopter drones to monitor the integrity of bridges.
“The idea is that bridges in the [United States] need to be inspected every year or so, and they usually want an engineer on site to make a visual inspection,” Tran said. “For some bridges they have sensors to measure the strain and tension as well. For a large bridge you have to close down the traffic and send out an engineer to actually do the inspection.”
Drones, Tran said, could take the human out of the equation.
“The idea is that instead you have a drone equipped with a camera to make the visual inspection,” she said. “And if there are sensors on the bridge you could also have the drone go up to the sensors, collect the data and then return with the information.”
Most of the theory for this project has already been verified, according to Xi, but applying the theory presents its own set of challenges.
“The main framework of the theory was already established, but it turns out that in many practical situations there is more to be developed,” Xi said. “For those parts of the process there is often not a lot of established work, so that is some of the theory that we work on: For example, we worked on making the drones work even when there is ‘noise,’ or interference.”
At Tufts, the SPARTN Lab is unique in the way that it connects undergraduate, graduate and Ph.D. students. The interactions, according to Khan, are what drives the lab toward new innovations.
“We start with some initial ideas, and then the process becomes very interactive,” Kahn said. “It is very interactive because these are open-ended problems and there are many ways to solve one of these problems. It is an evolution based off of interaction with students.”