As we wrote last week, we've been preparing a pilot run of 50 robots. This has been really useful for working out manufacturing and development issues, but we also used the robots for a demo at TED 2014! The last 10 days have been a real whirlwind, and now that the demo is over, I'm happy to show how much progress we've made in such a short time. There is a lot of stuff! We'll show some of it this week, some of it next week, and a bit more after TED releases their video of the talk! But first, a small swarm...
A few weeks ago, I got a call from Berkeley Professor and great friend of Dash Robotics, Bob Full, saying he would be doing a short TED talk in March, and would we want to have the robots featured. Yes! was the easy response. Of course, we had to think of something that would wow people. We figured 50 robots running at the audience and all over the stage would be our best bet. After a lot of planning, and a lot of work, we pulled off that demo on Tuesday. Things got really busy in the last week, here's what we've been up to.
The first thing we had to figure out was, how the heck do we trigger 50 robots all at once? Bluetooth won't let us do that, there's all kinds of crazy interference at these conferences, we don't want to risk infrared... We decided that a light trigger would be the best. The robots could sit in a dark box, waiting to see light. Once they do, they run. Of course it sounds simple, but in practice, for a demo that absolutely cannot fail, there is a lot to consider.
On March 8th, we only had 4 Dash robots in existence, and only 1 set of electronics, which only ran some very basic code. By March 18th, we would have 56 robots ready to demo, all with our electronics and code on them. March 9th, as we showed in our last blog post, was a successful frenzy of robot building, giving us our 56 robot chassis and powertrains. The electronics however, didn't come until March 11th. This was thanks to a heroic effort by our contract manufacturer, who managed to deliver 56 working boards (100% yield) on very short notice! This box was very exciting to open:
And ultimately led to this:
Each robot, with its own electronics, motor, battery, the works. Now we were ready to rock. We even gave them nametags!
Once we had the robots, we could start testing the demo. The first thing was to make sure they would run when triggered by light.
Here, we see Dwight triggering the robots with an iPhone flashlight. Works like a charm! But we realize that the robots don't really go very straight, and instead just do their own thing. Since Dash is equipped with a gyroscope, we can use the readings of the gyroscope to tell how fast he is turning, and then tell the motors to speed up or slow down accordingly. Some coding ensues, and we run some tests to see if we can't get them to straighten out a bit.
Here, Dwight is trying to keep track of which robots have the new code, and which have the old code. After a while, we got pretty good at developing systems for dealing with this. Once the new controller was on the robots, we could test them to see if they run straight. It took a few iterations, but eventually we were pretty happy with their performance.
Now that we had robots that ran straight, we needed a light box to release them from. Andrew actually had to learn how to sew to do this.
A big thanks to our friend Leanne for the tutorial :)
The lightbox comes together...
Made of black fabric and aluminum tubes, the box shields the robots from light, and also can be disassembled and brought on a plane.
It came time to test the swarm under our creation:
Not bad, but still a lot of work to be done! Next week, in "The Swarm Unleashed - Part 2," we'll show how we got from here to a demo-ready setup, the problems we faced, and what we did to overcome them. Thanks for reading!