Andrew Barry, a PhD from MIT’s Computer Science and Artificial Intelligence Lab (CSAIL) has built a drone with has the capability to detect obstacles far off and veer away from it to avoid collision.
Don't Miss: Incredible Pokemon Gifts
The developed drone can dip, dart, or dive through a forested area at a speed of about 30 miles per hour, and its obstacle-detection system enables it to avoid obstacles with the expertise of a piloted plane.
“Everyone is building drones these days, but nobody knows how to get them to stop running into things,” said Andrew Barry, who developed the system as part of his thesis with MIT professor Russ Tedrake. “Sensors like lidar are too heavy to put on small aircraft, and creating maps of the environment in advance isn’t practical. If we want drones that can fly quickly and navigate in the real world, we need better, faster algorithms.”
Barry’s new drone can run 20 times faster than those already existing in the market, and its stereo-vision algorithm helps it to map its surrounding and then detect objects in real-time. Its software operates at 120 frames per second, and it pulls data at a speed of 8.3 milliseconds per frame.
With a wingspan of 34 inches, the latest drone weighs only about a pound, and the over-the-shelf components that went into its making cost about $1,700. It has two cameras mounted on both wings, and its processors are not much different from what mobile phones have.
Understanding that existing drones use algorithms of photos that their cameras capture in order to search depth-field of multiple distances – a problem that limits the drone’s flight to 5-6 miles per hour, Barry’s drone was based on computations that go farther than that on a subset of algorithms.
“You don’t have to know about anything that’s closer or further than that,” Barry volunteered. “As you fly, you push that 10-meter horizon forward, and, as long as your first 10 meters are clear, you can build a full map of the world around you.”
But he is not resting yet, because the current method used to develop further speed for the drone could limit its drive due to the missing information it has from integrating data from its odometry and previous distances. However, Barry is still working to enable his drone work at several depths and in dense forests.
Buy Now: Sony PlaysStation VR In Stock Here
“Our current approach results in occasional incorrect estimates known as ‘drift,’” he saied. “As hardware advances allow for more complex computation, we will be able to search at multiple depths and therefore check and correct our estimates. This lets us make our algorithms more aggressive, even in environments with larger numbers of obstacles.”