Robotics researchers at MIT have developed a system that lets a toy-size airplane quickly figure out where it is inside a building as it is flying. The system allows the vehicle to swoop around corners and take tight turns, all on its own.
Outdoors, robotic vehicles can learn where they are by gathering GPS coordinates. But indoors, GPS doesn't work, so small, autonomous aircraft that can perform quick, aggressive maneuvers must rely on cameras set up around the ceiling that track the craft's every move. (Larger vehicles, such as ships and full-size airplanes, use bigger navigation systems that can calculate their own GPS coordinates).
Small, autonomous aircraft that don't use motion-tracking systems either move very slowly or crash into obstacles, reported IEEE Spectrum, an industry magazine.
The MIT group's plane is different. Researchers have to load its on-board computer with a 3D map of the room. Once it has that, it uses lasers and an IMU (a device for aircraft that measures the craft's velocity and orientation) to quickly estimate where it is in the 3D map. Watch the self-flying plane avoid walls and columns and navigate hallways:
Ultimately, the research group wants to make small, autonomous flying vehicles that can estimate their own positions and build maps of the world around them even without GPS, according to the group's website.
The MIT researchers, led by student Adam Bry, presented their research May 15 at the Institute of Electrical and Electronics Engineers' annual robotics meeting, where their study earned a nomination for best paper.
Though the MIT team's paper doesn't discuss future applications for their self-flying plane, other researchers say that such machines are a first step toward flying robots that can scope out buildings after a disaster.
Follow InnovationNewsDaily on Twitter @News_Innovation, or on Facebook.
