TED BlimpBot Report: Foiled by Air Conditioning!

Chris Anderson
The Long Tail

March 01, 2008

[This is a cross-post from my other blog at DIY Drones [ http://diydrones.com/ ], where among other things we've been developing an under-$100 autonomous blimp for aerial robotics contests]

Yesterday my UAV partner Jordi (a 21-year-old embedded programming whiz who just emigrated from Mexico) drove up with the BlimpBot to Monterey, where I was attending the TED conference [ http://www.ted.com/index.php ]. I had a three-minute slot to show how the prototype works. This was a pretty high-stake demo, since not only would there be 2,000 of the most influential people in the technology, entertainment and design (TED) worlds watching, but they included Al Gore in the FRONT ROW, Google's Sergey Brin and Larry Page and movie stars such as John Cusack and Goldie Hawn.

The bot worked great in the hotel room, and then we took it the the auditorium during a break to test it on the main stage. Yikes. We were getting IR interference from everything, from LCD screens to the bright stage lights, and our reception range dropped to something around three feet. Even worse, the air currents were overcoming the blimp's ability to fight them. So we gave up on the idea of a fixed IR beacon on the ground, and I decided to hold it in my hand to keep it near the blimp. Even then, the motors couldn't fight the currents well enough.
So we rushed back to our staging area (my hotel room) and Jordi updated the firmware to give more power to the motors even at the cost of battery life (this demo only had to run three minutes) . We tested it again in the hotel room, it worked fine, and then it was time to go.

When we got to auditorium and waited in the wings to go on, it was clear that something bad had happened in the firmware update. The vertical motor wasn't coming on at all sometimes and it wasn't clear why. Then Jordi realized that in changing the power settings,he'd also changed the timing of the loops, and we weren't resetting the motor controllers at the right time, which meant that the chance of them being working when needed was random (and low). We'd just been lucky in the hotel room, but clearly weren't now. Still, I crossed my fingers and went on, carrying the blimp.

Disaster! It turns out that one big thing had changed since our test run in the auditorium: 600 people had arrived. All that body heat had raised the temperature of the room, kicking in the air conditioning, which came out of huge ducts right over the stage. Basically I was under a raging waterfall of cold air, and the poor blimp sank right to the floor, its little vertical thruster completely overcome.

Lesson 1: Little blimps need still air
Lesson 2: If you can't find still air, you need WAY more powerful thrusters (which means more battery power, which means more weight, which probably means a bigger blimp)
Lesson 3: Don't update your firmware five minutes before you're going to fly an autonomous robot ten feet away from a former Vice President of the United States.
Lesson 4: Hey, it's a tech demo on stage, and they *always* go wrong--don't let it throw you. So I didn't. I just stood there holding the blimp, as you can see in the picture above, and went on with my talk and slides as planned. Points made, time limit met, applause gained. I looked a bit awkward, I'm sure (although hopefully not always as unhappy as I look above), but at least I got the sympathy vote! Now on to San Diego for Etech [ http://en.oreilly.com/et2008/public/schedule/detail/1607 ] on Tuesday, where we get to do it again for an hour in front of the smartest geeks in the world. So much for the sympathy vote ;-) Jordi's hard at work fixing the firmware problems, so fingers crossed...

[Photo credit: Red Maxwell]