Fleets of Drones

Fw:Thinking
140 followers
12 Video Views·Dec 12, 2022
Need a patriotic display to hang in midair?
Send in the drones!
If you watched the 2017 Super Bowl halftime show,
you might remember the dazzling image of the American flag hanging above the stadium.
Each light making up that image came from a drone
flying in a specific configuration to create the effect.
So, what's going on here?
The drones are from Intel,
and they're calling the nighttime display drones Shooting Stars.
They fly in unison, creating complex patterns in the sky through precise positioning
and an LED lighting system capable of replicating four billion different colors.
They've even partnered with Disney to create special light shows at theme parks.
It's an impressive display of technology and computer science.
Your typical quadcopter drone takes orders from a human operating a remote control.
If you wanted a group of drones to fly in formation,
you needed several drone pilots with steady hands to pull it off.
Intel's approach was to design software giving a single operator full control over a fleet of drones.
This isn't quite swarm intelligence because the drones aren't communicating with one another.
The operator's computer sends out instructions to each drone in the swarm.
Those instructions tell the drone where it needs to be
and gives it the instructions to follow the specific sequence of moves.
That means you need to plot out the full sequence on the ground
well before drones take to the sky.
You have to make sure that the courses you plot won't bring two drones too close together
or else you'll have a collision.
Fortunately, Intel Shooting Star drones are made of lightweight materials like foam and plastic,
and their rotors are inside protective cages,
so even if something unexpected happens, the chances for real catastrophe are low.
The Shooting Star performances are undeniably impressive,
and reducing the number of operators to one is incredible,
but the next step involves drones that can work together
to carry out instructions from a single operator.
The Intel approach has each drone listening only to the computer program,
but swarm robots also listen to each other, and that brings us over to MIT.
Imagine designing an algorithm that lets a drone sense its environment
and communicate with fellow drones while trying to follow instructions from an operator.
You might have a seemingly simple formation in mind,
like a triangle made up of three drones,
but all the variables in play make this apparently simple request much more challenging.
The MIT approach is to use decentralized algorithms.
That means each robot has just a part of the overall information,
such as who its neighbors should be and how its position should relate to those neighbors.
It's like being in a dance line and keeping an eye on the people to your left and right
so that your timing is perfect.
I, uh, was a theater kid.
The Intel approach is a centralized algorithm
in which a master computer sends out instructions to a bunch of essentially dumb drones.
The MIT strategy is to distribute that decision making
throughout the fleet of drones to create a true swarm.
Neither solution is inherently superior.
The best one really depends upon what you plan to do with the drones.
I think swarm technology is particularly cool
because it introduces robots to the realm of group decision making.
Of course, that might mean we're ultimately headed into a future of robot committee meetings
chaired by actual robots.
But I have something for you guys to think about.
I want you guys to come up with some creative, positive uses for swarm robotic intelligence.
Think of stuff like allowing cars to chat with each other so that they can avoid collisions.
Let me know your ideas in the comments below.
Thanks to Toyota for sponsoring this show.
If you enjoyed this episode, hit the like button down there, would ya?
And be sure to join the super awesome forward thinking think tank by subscribing to the channel.
