WPI students building autonomous security robot for U.S. Air Force

Advertisement

A team of undergraduate students at Worcester Polytechnic Institute (WPI) in Worcester, Massachusetts is attempting to turn a wheelchair into an autonomous security robot that could help protect unmanned installations for the U.S. Air Force.

The students are building this proof-of-concept robot using a tracked all-terrain wheelchair (trackchair)— provided by a company called Action Trackchair Inc.—and about $8,000 for supplies.

According to WPI, the military wants to have robots at unmanned bases that can respond to and investigate intrusion alarms and other warnings within two minutes. The robots would be used to safeguard thousands of unmanned bases, such as missile silos, around the world.

For about a decade, the Air Force has been offering student-based, high-tech competitions. For the first time last year, the Air Force ran the autonomous security robot challenge, which saw three teams compete against each other. All three teams made advances during the challenge, but they each ran into an obstacle they couldn’t overcome, so no winner was named.

This year, the WPI team was the only one to accept this specific challenge, so there is no competitive aspect to the event, but James Gilland, a research team manager at Ohio Aerospace Institute—the entity that is running the challenge for the Air Force Research Laboratory—says that the Air Force is “still looking forward to seeing what kinds of innovations the WPI students can develop.”

“The path for the program is not necessarily for the students to hand over a finished product that the Air Force will use as is,” Gilland says.

“For this challenge, there are so many aspects to the functioning of a security robot that there are multiple goals in this one problem. Even if you’re the only one in the challenge, you’re winning if you’re advancing the technology in a new and unique way.”

The robot, equipped with off-the-shelf sensors, cameras, and video equipment, has to be able to unplug itself from a charging base station, then quickly and autonomously navigate to the spot where the intrusion was detected.

Once at the site, the robot must be able to transmit photos and video of what’s happening at the location—whether it is an actual human intrusion or another issue—to an Air Force base, where humans will decide if security guards need to be dispatched.

To meet these requirements, the WPI team is building algorithms for autonomy, machine learning–based vision processing, and navigation. The team is also fabricating a sensor suite, a base charging station, and a communication network.

The team has also installed sensors and cameras, set up a controller, and has been working on teleoperation controls.

According to one of the students on the WPI team, Marissa Bennett, one of the challenges that the team is currently facing is how to keep the robot moving at its top speed with the added weight of the sensors and cameras and the extra power those components will consume.

The most significant challenge, though, might be equipping the robot to work in a wide range of temperatures and environments.

“The robot itself has to be thermally equipped to maintain a lot of different electronics onboard,” Bennett says.

“It needs to keep them at a stable operating temperature, whether it’s below zero or in extreme heat, so the robot can function no matter where it is. For instance, we’re coming up with a way for the robot to cool itself by pushing air through itself as it’s moving.”

Gilland adds that he is hopeful that the WPI team will find creative ways to make the robot see, sense its surroundings, maneuver, and guide itself. Gilland is also excited to see how the team will make sure that the robot does not tip over in difficult terrain.

“The part that will be useful for us is to see which algorithms work better than what the Air Force has been working on,” Gilland says.

This spring, the students are expected to demonstrate their security robot at Wright-Patterson Air Force Base near Dayton, Ohio.