SHARE

Stroll by a strip mall arcade or the local Dave & Buster’s, look behind the noisy kids playing Dance Dance Revolution, and you’ll likely spot an air hockey table. Like Pac-Man and the maddening claw game, air hockey remains unchanged and everlasting. Two facts seem to endear us to the floating puck: 1) everyone thinks they’re good at the game but 2) nobody knows for sure. Nowhere in the sports landscape are so many goals scored upon oneself. A 6-0 victory in one game is reversed in the next battle, thanks entirely to Lady Luck. But when you compete against the Air Hockey Bot 1000 (AHB-1000), a career once dictated by fickle fortune can finally be tested against formulaic consistency.

The second coming of air hockey was born from an unlikely source. Nuvation is a company that offers a variety of electronic design services, helping corporations develop everything from MP3 players to space shuttle components. The San Jose-based company was originally contracted by Freescale Semiconductor to create an air hockey robot, to demonstrate the difference in speed between its 8-bit and 32-bit microprocessors. Two months later, with just three employees dedicated to the project, the first prototype of the AHB was demonstrated at a technology forum sponsored by Freescale. The response has Nuvation considering commercialization.

“There’s been so much excitement and interest about it,” said principal design engineer Mohan Gurunathan. “We’re talking about various possibilities, expanding into arcades or amusement parks and building other games. The response has been overwhelming and we’re still trying to get our bearing on where to go with this.”

The technology within the robot is as surprisingly simple as the game it dominates. The system can be broken down into three components: the vision system, the intelligence, and the actual robot. The vision system consists of an off-the-shelf machine camera (a model often used for assembly line inspection) mounted eight feet above the table. The camera is encircled by a custom array of lights directed onto the playing surface. A professional-grade puck is then covered with reflective tape, similar to that often seen on backpacks or running shoes. During play, the vision system locates the puck, defined as the center of the bright light, 100 times per second. That location is relayed to the computer, where an algorithm determines the expected path of the puck. The engineers developed the algorithm based on theoretical calculations, taking into account spin and energy loss in conjunction with a series of controlled tests. An off-the-shelf industrial robot (designed to paint, seal, and sand parts) is then instructed to move to the expected location. According to Nuvation, the robot can’t move as fast as the human hand (just 4 meters per second), so save your excuses.

The robot’s offensive prowess is (currently) a bit less sophisticated than its defensive game. Once the computer determines the puck is no longer a scoring threat, the robot flips to swat mode. The swat maneuver is arbitrarily directed away from its home goal and does not yet calculate an optimal path or a complicated bank shot. To win, the robot instead relies on a favorite sports mantra of coaches nationwide — the best offense is a good defense. During the Freescale forum, the cliché held true.

“It scored on its own every now and then,” said Gurunathan. “But more often people would get so frustrated trying to score that they would score on themselves because they stopped playing any defense.”

Ah, but even robots must contend with the agony of an own-goal. While the robot can block anything in sight, a puck which travels beneath the robot arm, and therefore out of sight from the camera, can confuse the computer, causing it to accidentally score on itself.

Gurunathan surveyed previous attempts at air hockey robots and is confident the AHB is the fastest and most capable to date. Easily accomplishing its original intent of showing differences in processors (8-bit was much slower), development focus has shifted towards creating a market-ready product. Improving the user interface is paramount; installing a more imposing offensive gameplan should be easy. Even the behind-the-arm flaw will likely be fixed before anyone pops in four quarters for a dose of humility. With each enhancement, Gurunathan will have a group of semi-pro air hockey-ers in the San Francisco area push the computer to its limits.

Semi-pro, eh? I’ll take my chances.