Expensive industrial robots have little in common with robotic toys, right? Well, don’t tell that to Aaron Rasmussen and Brian Mullins, cofounders of USMechatronics. While taking a weekend break from their day jobs designing robotic cells and other complex electromechanical systems, they decided to teach a KUKA KR16 robot to play tennis and, for a few more laughs, do some sword fighting.
Rather than simply program the robot as they would for one of their commercial jobs, they decided to create a wireless control system based on the Wiimote, the motion-sensing remote control unit for Nintendo’s Wii game console. Users simply swing the remote through the air and the nearby robot makes the corresponding swing of the tennis racket or sword.
“The idea was to take one industrial robot, add a laptop talking to a Wiimote, strap on a tennis racket, have it follow the swings the user makes, and do it all in a few hours on a Saturday so we could get back to our busy schedules,” Rasmussen writes on a Web page, describing the project’s software development and hardware.
Rasmussen says the germ for this “WiiBot” project was all the buzz about “cool Wiimote hacks” that allow it to control things other than Nintendo’s game console. “The Wiimote communicates via Bluetooth, so it can be made to talk to just about any notebook computer,” he says. And though it costs only about $40, the Wiimote contains an array of solid-state accelerometers that can sense translational and rotational movement. “I think of it as providing accelerometer access for the masses,” says Rasmussen.
Remarkably, Rasmussen was able to get the WiiBot to “start swinging” in just a couple of hours. He used free software to get the Wiimote to talk to his laptop computer, and he wrote a program to translate the user’s swings with the Wiimote into one of six preset swings Mullins created using the 6D mouse on KUKA’s robot controllers.
At first, Rasmussen tried to use free game controller emulation software called GlovePIE to turn the minimum and maximum values from the Wiimote’s accelerometers into move values the KUKA controller could understand. But he had some trouble getting enough differentiation between different swings, and GlovePIE didn’t allow the robot to play tennis very well. “So I sucked it up and wrote some simple pattern recognition software of my own,” Rasmussen recalls, explaining that his software matches the user’s Wiimote swings to the robot’s preset swings using a simple mean-error algorithm. “I didn’t bother to use mean squared error. That’s how lazy I was,” he jokes.
Turning serious, though, Rasmussen stresses this robotic mash-up doesn’t represent the kind of work he and Mullins do commercially. “It’s important to note that Wiimote doesn’t replace the KUKA robot’s own controller,” he says. Nor does the idea of roughly matching the Wiimote movements to preset robot movements achieve the accuracy or precision required by most industrial robot applications. “The wireless remote functions more like a human-machine interface,” Rasmussen says. Finally, the idea of controlling a robot wirelessly in an industrial setting would send many engineers running toward the shop floor exits.
Still, the WiiBot provided Rasmussen and Mullins with several hours of amusement. “We meant to make a video of the robot that Saturday, but we played with it all day and ran out of time,” Rasmussen says. They ended up filming the video a few days later, and it quickly drew so much attention and enough interest on the Internet to crash USMechatronics’ servers.
And the WiiBot project may even have implications beyond fun and games. There’s a big push right now to make robots more accessible to end-users, according to Kevin Kozuszek, director of marketing for KUKA. “In that sense, the Wiimote is a great learning tool,” he says.