Why walk when you can fly The U.S. Navy is interested in a scenario where flying robots replace humans for inspecting a ship's tanks and the irregular areas around them for corrosion.
With funding from the Navy Engineering Education Consortium (NEEC), Professor Dan Stilwell and a team of graduate and undergraduate students in the Autonomous Systems and Controls Laboratory are building autonomous drones to do the job.
Because ships reside in saltwater, they tend to rust, Stilwell explains, and the Navy must check the tanks regularly to make sure the ships are safe. "It's time consuming and expensive," says Stilwell. "We're building a robot to fly into the unknown area and check for corrosion."
Ph.D. candidate Matthew Bailey is working on a ducted-fan unmanned aerial vehicle. This is a flying vehicle with a single propeller that operates inside a duct. This configuration maximizes propulsive efficiency and increases endurance. Because the ducted-fan vehicle is expensive, a team of undergraduates is doing some initial work with a small fleet of more affordable quadcopters.
Jyssica Baehr, Randolph Peterson, and John McDouall discuss the next steps for their project.
The undergraduate team consists of four seniors, two juniors, and five sophomores: Jyssica Baehr, Xin Gan, Griffon Jarmin, Ryan McCall, John McDouall, Marc Murphy, Randolph Peterson, Peter Quan, James Reed, Zhun Shi, and Ryan Willard. They have divided into four smaller groups, each working on its own quadcopter.
Controlling quadcopters indoors
The students are working to control the quadcopters when flying insidea task complicated by the dynamics of the airflow generated by the copter itself, Stilwell explains. "The copter moves a lot of air in order to generate lift, and if it gets up close to something indoors the motion of the air can change and the copter might just fall to the ground."
The student' first task was choosing and building the quadcopters. They selected their quadcopters because of the open source software used for the flight controllers. "If we want to reprogram some of the behavior of the flight controller, we can do that," Reed explains. "With a proprietary system, it would be a black box and we would have to make compromises."
By making open source controllers a priority, they chose a quadcopter that was more difficult to build and fly than other commercial varieties. "The instructions to build these are pretty horrific" says Peterson. However, the team does not regret the challenge.
Jarmin explains that "there's not much assembly instruction, because it's up to you what to do with it. It leaves you to act just like an engineer. Which is what we're trying to be." Also, he continues, "building is what allowed me to understand every aspect of how it works. It's how you learn what you need for the control algorithms."
Flying along a wall
Their next challenge is to make the drones approach a wall and fly along it. Flying the quadcopters inside is easy, according to Reed, "as long as you don't get close to walls."
Shi explains that features like following along a wall aren't available for commercial quadcopters."We want to add new features to what is available," he says.
According to McDouall, "a big part of the project is going to be stability in unpredictable conditions."
Peterson continues, "when you get close to the wall, the air currents change. We have to account for that and maintain stability."
The students are enjoying the hands-on experience and the freedom they have to solve the problem. "Our goals are dynamic. If we stumble on something cool, we have free range to see where it goes," says Jarmin.
Eventually, these drones will also be required to explore and map the area it is checking for corrosion. The mapping will also help compensate for the indoor airflow problem, according to Stilwell. "As we build a map, we can anticipate the airflow," he explains, "so we can fly in very complicated areas."