The team behind the research, MIT's Robust Robotics Group, previously developed helicopter micro-UAVs which have won challenges set by Unmanned Vehicle Systems International (AUVSI). In recent years those challenges have focused on indoor navigation without the benefit of GPS navigation.
But with its latest UAV, the Robust Robotics Group has decided to up the ante by opting for an autonomous fixed-wing aircraft: two taxing technological problems to overcome for indoor flight. The advantages of an autonomous (i.e. self-governing) UAV are obvious, but its fixed wings are a boon, too.
“The reason that we switched from the helicopter to the fixed-wing vehicle is that the fixed-wing vehicle is a more complicated and interesting problem, but also that it has a much longer flight time,” said Nick Roy, head of the Robust Robotics Group. “The helicopter is working very hard just to keep itself in the air, and we wanted to be able to fly longer distances for longer periods of time.”
To make the switch from a helicopter to fixed-wing aircraft, the Robust Robotics Group designed and built their aircraft from scratch with the assistance of Prof. Mark Drela of MIT's Department of Aeronautics and Astronautics (AeroAstro). “He’s a guy who can design you a complete airplane in 10 minutes,” said AeroAstro grad student Adam Bry. “He probably doesn’t remember that he did it.”
The Group developed a small airplane with a wingspan of 2 meters (6.5 ft) weighing just 2 kg (4.4 lb). The plane's wings are deliberately short and broad for its length, which the team says allows the UAV to fly at low speeds and turn tightly, while being able to carry some additional weight (i.e. its electronics).
The drawback is that the UAV cannot hover. There is a minimum speed that must be met at all times to avoid stalling. So, in addition to sensing its environment and planning a safe route, the machine must also check that the routes are “dynamically feasible,” or that they can actually be flown.
The UAV's eyes and ears are an onboard laser range-scanner and an inertial measurement unit. The range-scanner operates continuously, but is only capable of grabbing a 2D picture of obstacles and free space in its environment: insufficient information to ascertain its own location in the space. But with the help of specially-developed algorithms (really, the crux of the technology), the onboard processor (a modest Intel Atom) is able to work out the UAV's position by analyzing successive scans, and is helped along with the additional information provided by the inertia unit.
The catch to all this is, at present, the aircraft needs to be pre-fed a map of the environment in which it is to fly. It's a luxury not afforded to the helicopters that typically compete in AUVSI's challenges. Building maps, literally on the fly, is a hurdle to overcome, but even so, the team's technical accomplishments are not to be sniffed at.
The video below includes footage of the UAV in flight, perhaps most impressively through a multi-story car park where the vertical height in the space is sometimes limited to 2.5 m (8.2 ft), leaving the aircraft very little margin for error when making sharp turns.
Adam Bry is lead author of two research papers related to the UAV's development, Rapidly-exploring Random Belief Trees for Motion Planning Under Uncertainty (PDF), and State Estimation for Aggressive Flight in GPS-Denied Environments Using Onboard Sensing (PDF). The research is not directly related to MIT's research into Goshawk-inspired agile UAVs.
A video of flight tests can be seen below.