With just a week left in the Pitch Your Prototype challenge, Jose Julio’s entry is more lighthearted than some of the more practical or conceptual submissions so far — though it’s still technologically impressive. Julio’s daughter enjoys air hockey, so he wondered whether he could create a robot that could hold its own against a human player.
“I wanted something easily transportable and comfortable to play, but at the same time, easy to replicate and open source,” Julio wrote. “After some weeks scratching my head and problem solving, I can proudly state: It’s air hockey time!”
Julio decided to combine standard RepRap 3D printer components — stepper motors, drivers, belts, bearings, rods and an Arduino Mega — with a video detection system. It was a cheap way, he said, to give the automated opponent two axes of control with minimal weight.
He replaced the X-axis rods on the RepRap with carbon tubes, which worked with PLA-printed bushings and made the system lighter.
To detect the location of the puck, Julio used a PlayStation Eye connected to a PC running a vision system he wrote using OpenCV libraries. The computer records a video of the input, which Julio said was useful for debugging, and then sends the location data to the Arduino over a serial connection at a rate of 60hz.
Finally, he wrote a trajectory prediction system and the robot’s air hockey strategy. Two captured frames since the last impact give enough data to plot the trajectory of the puck at any moment, and that data feeds into the artificial intelligence that decides whether to defend, attack, or try a combination of tactics. Conveniently, the insulated strategy subsystem makes it easy to adapt the system’s AI without reinventing the wheel with coordinate and trajectory analysis.
“A nice feature of this project is that the strategy subsystem is fully insulated and is very easy to modify and reprogram, so you can develop your own strategy algorithms from the complexities of motor control, vision system and prediction code,” Julio wrote.
The Pitch Your Prototype Challenge is a collaboration between Make: and Cornell University, with the goal of digging up promising ideas from the Maker community. The individual or team that wins the challenge will be awarded $5,000 and have the opportunity to appear onstage at MakerCon New York. Visit this site to enter your own prototype or vote for a winner.
You can read about more of Julio’s projects on his blog, JJRobots.