Two Sigma Investments – Human vs Machine

Published on Apr 21, 2016

Each year we like to challenge ourselves with a hard, but fun, engineering problem. It always involves programming artificial intelligence systems to battle in a head-to-head competition. In previous years we’ve built AIs to play computer games, card games, and shuffleboard against our employees and each other. 2016’s epic battle saw humans and AIs squaring off across an air hockey table. That’s right—we programmed AIs to play air hockey. How? You might wonder. And who won? Check out this video to find out, and witness the complete story of the 2016 TS Cup.

Two Sigma Investments

0:15ladies and gentlemen
0:23we have a history of programming competitions you want something that was
0:29exciting for an audience to watch and that’s where the idea for the robotic
0:33air hockey game in two players in this is probably the most exciting
0:35competition that we put together so far
0:39the air hockey table we built this contest was developed as a iterative
0:43design we worked with some outside a robotics experts to design a new system
0:48that was inspired by the cable cameras used at football games that we close the
0:53loop with a vision system that was designed by the other team members who
0:57uses two of these PlayStation Eye cameras each mounted on one of these
1:01brackets here through these cameras we all get two different pictures of the
1:05table from the full blue table we can then find the Green Park and the Red
1:11mallet using computer vision the position of the puck in the mallet is
1:14then relayed to the contestants by the game server the challenge was in the AI
1:19teams to take the vision that we provide to them and generate the motor commands
1:23to be able to eat the other a eyes as well as a human from the human rockets
1:57when you give people some time away from their deaths to be creative
2:04a lot of innovation can happen when they’re working on these activities are
2:07using programming languages they’d never used before they’re using libraries that
2:11they’ve never experimented with before I am Python stuff we gave me a chance to
2:15practice using some areas of Python that I don’t normally use and the chance to
2:19be a little bit experimental in a trust fund
3:10to play conservatively look at the position of the pocket and the
3:21directions that the puck is moving in and predict where it’s going to be one
3:25thing I learned it was simple as probably always better you can
3:28definitely going to this kind of problem and come up with all sorts of crazy
3:32solutions that will take a long time to implement and debug just try the
3:36simplest things first
3:37essentially tried to make a play like I would play air hockey so I was
3:41definitely going for speed and general being in the right area at the right
3:45time rather than a curse
Two Sigma Investments

4:02surprise was that instead of playing manually with my hands I got to meet up
4:08with an oculus headset and had to control using a touch screen so I got
4:13the exact same interface but the robot
4:19definitely a celebration of engineer to Sigma came out victorious because it’s
4:28an opportunity for two sigma employees to really get involved in new and
4:33cutting-edge technologies really the medium is the message I try something
4:37different to really energy next year I hope we do this again and I will
4:40absolutely be back to try some of the things I wanted to try I’m really

Two Sigma Investments

Two Sigma Investments - Human vs Machine
Source: Pixabay