Drones are pretty cool devices – or should we say, toys, to own, especially when you have a whole boatload of money and love to see these machines flit their way through the air, snapping quality photos from above and the ilk. Well, they are not too great when it comes to avoiding obstacles that are in front of it, which would definitely make for a heart-breaking moment should your drone ram into an obstacle while going at full speed ahead. How about a drone that is so agile, it can avoid the strokes of a fencer?
This is exactly what Ross Allen and Marco Pavone, researchers at Stanford University’s Department of Aeronautics and Astronautics, have set out to achieve – by programming a drone obstacle avoidance system through a rigorous training method that will require the point of a fencer’s sword. Apart from utilizing a system that can see the incoming foil, there will also be aspects of machine learning thrown into the mix, in order for the drones down the road to be more agile and versatile.
At this point in time, data received by the drone is taken from an external camera, but such systems are being explored by major drone manufacturers such as DJI.
Filed in . Read more about Drones. Source: sploid.gizmodo