Clik here to view.
Image: MIT
Your average drone Image may be NSFW.
Clik here to view.
is dumb, but they are getting smarter. Smart features such as auto-follow and waypoints are being integrated into today’s consumer drones, but it’s still tricky to teach drones how to avoid obstacles. While some drones are designed to bump into things, researchers at MIT are developing planning algorithms that’ll keep drones out of harms way.
MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) recently demonstrated their planning algorithms on drones in a simulated “forest.” With these algorithms in place, drones were able to plan and perform agile maneuvers on a dime. Stunts such as donuts and figure-eights were performed in a matter of seconds as the drone zoomed past the “forest” of PVC pipe and interwoven string.
So how exactly do these algorithms work, and what’s the technology behind it? Putting it in simple terms, the algorithm divides space into “obstacle-free regions” and then stitches these regions of space together to form one obstacle-free route. More traditional algorithms count the number of obstacle faces, but this can result to dwindled performance as the system encounters more obstacles to track. A real-life setting is bound to have more obstacles than a lab setting. On the other hand, CSAIL’s new algorithm could handle densely packed environments simply by counting the number of open routes (far less than the number of obstacles present).
“Rather than plan paths based on the number of obstacles in the environment, it’s much more manageable to look at the inverse: the segments of space that are ‘free’ for the drone to travel through,”
explains recent graduate Benoit Landry.
For now, MIT’s drones are being tested inside laboratory settings equipped with Vicon trackers and idle obstacles. Realistic applications such as inspection and delivery take place in open environments that are not accoutered with fancy tracking systems and non-dynamic obstacles. It’s going to take a bit more research to develop your next-generation construction inspectors and delivery men, but MIT has taken a great step into making that a reality.
Source: MIT
The post How Planning Algorithms Will Make Better Drones appeared first on SimpleBotics - Covering The Evolving World Of Robotics And Drones.