MIT ‘self-Flying’ Drone Can Avoid Obstacles At 30 Mph
Imagine a day in the not so distant future, companies (and nerds) operate drones with autopilot mode that can autonomously avoid trees, buildings and other obstacles in the air. Now, at MIT, there’s a drone that is a step ahead of other drones. One researcher at the Massachusetts Institute of Technology’s Computer Science and Artificial Intelligence Lab has developed a drone that can ‘self-fly’ at speed of up to 30 miles per hour.
In an interview published at Massachusetts Institute of Technology website, CSAIL PhD student Andrew Barry said everyone is building drones — and he’s telling the truth. From search giant Google, to social network Facebook and retail giant Amazon. They’re all testing drones that can improve their businesses. Google and Facebook are building drones to improve internet coverage, especially in far-flung areas, while Amazon wants to generate more revenue by developing a drone that can deliver goods to customers.
These drones are flashy, but for Barry, “nobody knows how to get them to stop running into things.”
Again, he’s preaching the truth. Back in June, the U.S. Federal Aviation Administration said regulators will finalize rules allowing widespread use of these flying computers by end of June next year — and their main goal is to protect people. Regulators want all drones to fly safely and avoid accidents that could harm a person, or a private property.
Now, Barry has created a software that can help drones fly more safely — and this drone rely on software to avoid obstacles.
MIT Drone that can ‘fly itself’
It’s like a drone with ‘self-driving’ feature, like Google’s driverless car — Barry’s drone is equipped with computer algorithm that can detect obstacles by building a map of surroundings in real time. Barry claims the new self-flying drone’s software is 20 times faster now, and its mapping tech can extract information at a speed of 8.3 milliseconds per frame, at 120 frames per second.
The good news, the flight code of the drone is available online at Github.
The main aim is to keep drones from hitting trees and other obstacles, while keeping speed.