Meet the GapFlyt. It’s a bio-inspired quadcopter, built by researchers at the University of Maryland (UMD), that uses only onboard sensing and a monocular camera to maneuver through small gaps.
The UMD team saw an issue with the traditional tactic of rendering a 3D environment for quadcopters and aerial robots. Designers today employ traditional computer algorithms for perceptual capabilities, which construct planning tasks that lead to autonomous behavior.
However, according to UMD researchers, “These methods are inefficient as they are not task driven and such methodologies are not utilized by flying insects and birds. Such agents have been solving the problem of navigation and complex control for ages without the need to build a 3D map and are highly task driven.”
The UMD strategy takes a minimalist approach—instead of reconstructing a 3D model of the scene, they use only the essentials to sufficiently complete the task at hand.
The quadcopter utilizes an optical-flow technique and visual servoing to maneuver through small gaps. As it captures images at certain frequencies while moving, it looks for similar features in the frames, and adjusts its relative location. By detecting how far certain objects moved, the drone calculates the gap’s edges.
The proposed framework was tested on a Bebop 2, a modified hobby quadcopter. “The Bebop 2 is equipped with a front facing camera, a 9-axis IMU, and a downward facing optical flow sensor coupled with a sonar,” according to the researchers.
After 150 trials with some gaps consisting of just a 5-cm tolerance, the drone recorded an 85 percent success rate at 2 m/s.
“To our knowledge, this is the first paper which addresses the problem of gap detection of an unknown shape and location with a monocular camera and onboard sensing. As a parting thought, IMU data can be coupled with the monocular camera to get a scale of the window and plan for aggressive maneuvers,” concludes the UMD team.
To learn more, read the article, “GapFlyt: Active Vision Based Minimalist Structure-less Gap Detection For Quadrotor Flight,” published in IEEE Robotics And Automation Letters.
Filed Under: M2M (machine to machine)