Game time: Robotics students showcase robots that drive, fly, and play Jenga
At the Johns Hopkins Whiting School of Engineering, you’ll find robots that assist in surgeries, help with search and rescue missions, explore outer space, and play popular games like Connect 4 and Jenga.
Last week, students in the graduate-level Robot Systems Programming course showcased robotics projects in demonstrations at the Homewood campus. Students designed, built, and programmed robot systems that can perform a variety of tasks, including self-driving cars and flying drones. Even the game-playing robots showcased skills that have important applications for modern technology.
For example, Chia-Hung Lin, a master’s student in robotics, programmed a robotic arm to play the block-stacking game Jenga—a game that requires skills that also apply to industrial robotics, says Lin.
“There are many jobs that will require a robot to do the same thing repetitively. Take a welding job—a robot will need to perform the welding job the same way every time. These are jobs that require coordination, dexterity and strategy—just like Jenga,” Lin said.
Lin’s robot first visually scans the position of the Jenga tower with a camera and a laser rangefinder. Then, it touches the individual Jenga game blocks with a delicate force sensor, determines the best move, and executes it—all in under 60 seconds. One of the most challenging aspects of his design was giving the robot the ability to change its grip on a game block—as humans can easily do—when executing moves required by the game. Lin’s solution? He programmed the robot to drop the block into a special parts feeder that Lin designed, from which the robot can pick up the block, now perfectly positioned in its gripper, to be placed on the top of the Jenga tower.
Prasad Vagdargi and Aalap Shah, both first-year graduate students in robotics, merged their backgrounds in mechanical engineering and computer science, respectively, to build a robot that plays (and always wins) the game Connect 4.
“We built a computer vision algorithm to understand the state of the board and detect certain colors so the robot can recognize the different game pieces. Then, the artificial intelligence component helps the robot plan the best move. Finally, we had to figure out the mechanics of getting the robot arm to execute the correct move,” said Shah.
The pair’s biggest challenge? “Everything ran in simulation, but nothing seemed to run in reality,” said Vagdargi. But setbacks didn’t slow the team. During the final demo, the clever robot beat Shah in just a few minutes.
According to Louis Whitcomb, professor of mechanical engineering who created and has taught the course for five years, this year’s cohort is the largest and most diverse he has ever had. The class included both graduate and undergraduate students from mechanical engineering as well as the Whiting School’s robotics master’s program.