MIT Student Creates an Advanced Autonomous Algorithm for UAVs

It looks like drones are indeed becoming more and more impressive as the years go by, and I’m not just talking about their speed and range. We’ve recently heard about a PhD student named Andrew Barry, who’s doing some research over at MIT’s Computer Science and Artificial Intelligence Lab. He has created an advanced detection algorithm that allows unmanned aerial vehicles to identify and avoid obstacles without any help from a human pilot. Even more impressive is the fact that this algorithm enables drones to avoid objects even if they’re flying through unfamiliar locations. Apparently, conventional algorithms are just too slow to keep up with the vehicles’ speeds, while advanced laser systems are not exactly practical because of their weight.

Andrew tested his creation on a small drone that was put together using off-the-shelf components. The device weighed less than a pound and had a wingspan of 34 inches. Moreover, it featured cameras and quad-core CPUs on each wing.

Typical  autonomous algorithms usually scan for obstacles at multiple distances such as one meter, two meters or three meters away. Barry’s program scans for objects that are 10 meters away, which is probably why it is so successful. During the testing session, the drone managed to fly through a forested area at a speed of 30 mph while avoiding every single obstacle along its path. The algorithm’s creator explained:

“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.”

Fortunately, the algorithm is open source, which means that you can actually implement it into your own UAV if you have the necessary equipment and knowledge. Have a look at the following video if you want to see Andrew’s drone in action.