This smart, agile MIT robot plays a mean game of Jenga
Researchers create a robot that combines vision and touch to remove Jenga blocks without toppling the wooden tower.
Robots are already master chess players and can solve a Rubik's cube in record time. So why not set them loose on the game of Jenga?
MIT researchers have developed a new kind of robot that's equipped with a soft-pronged gripper, a force-sensing wrist cuff, and an external camera -- all of which it uses to play a mean game of Jenga.
In Jenga, players take turns removing one block at a time from a tower constructed of 54 blocks. Each block removed is then replaced on top of the tower, which ends up making a more unstable structure. Whoever topples the tower by taking out a block first loses the game.
Using machine learning, MIT's new robot can look for the ideal block to remove. It does this by carefully pushing against a block, then taking in visual and tactile feedback from its camera and cuff and comparing the measurements to moves it already made to "learn" the ways of Jenga.
From the data collected, the robot then weighs the possible outcomes of different moves -- specifically, whether a block in a certain configuration, and pushed with a certain amount of force, was successfully extracted.
Using its robotic arm, the robot then makes a solitary, slow movement to take out the block without tipping over the tower... again and again. Details of the Jenga-playing robot are published in the journal Science Robotics.
Alberto Rodriguez, an MIT mechanical engineering professor, says it's the combination of vision and touch that lets the robot carry out tasks so well.
"Unlike in more purely cognitive tasks or games such as chess or Go, playing the game of Jenga also requires mastery of physical skills such as probing, pushing, pulling, placing and aligning pieces," Rodriguez said in a statement.
Researchers hope the robot's tactile learning system can be used in other tasks like separating recyclable objects from landfill trash and assembling consumer products.
"In a cellphone assembly line, in almost every single step, the feeling of a snap-fit, or a threaded screw, is coming from force and touch rather than vision," Rodriguez said. "Learning models for those actions is prime real-estate for this kind of technology."
Culture: Your hub for everything from film and television to music, comics, toys and sports.
My week with Aibo: What it's like to live with Sony's robot dog.