"Robots are klutzes, " says Ken Goldberg, an engineer AI expert at the University of California, Berkeley. A computer can easily defeat a human grandmaster at the game of chess by coming up with better moves. Yet a robot has trouble picking up an actual chess piece.

Though computers have advanced by leaps and bounds since 1980s, babies and kids still beat machines at certain types of tasks. 

The first task. Robots face three challenges in grabbing an object. Number one is the ability to locate an object. Even with advanced cameras and sensors that measure distance, robots still get confused by anything "shiny or transparent, " Goldberg notes. The second challenge is control. A robot's cameras and sensors won't always be in perfect sync（同步）with its moving "hand". And physics poses the final challenge. To grasp something, you must understand how that object could shift when you touch it. Physics predicts that motion. But on small scales, this can be unpredictable, for very tiny bumps on the floor or the object may change the motion（运动）.

Despite these challenges, humans grasp things all the time. Millions of years of evolution provided brains and bodies with ways to adapt. To help robots learn "robust（强有力的）grips"，Goldberg's team set up a virtual world DexNet where the AI model receive training. The DexNet contains more than 1, 600 different virtual 3-D objects and five million different ways to grab them. To be more like the real world, the team threw in some randomness. For each grasp, they shifted either the object or grabber just a little. After completing the training, a robot can figure out its own robust grasp for a real-world object it has never seen before.

Thanks to research like this, robots are getting less of "the klutze". Currently robots have trouble with other tasks including getting around the world, understanding people and thinking of new ideas. With engineers'work and efforts, it's expected that someone will design a graceful robot or even an AI model with common sense. "For now, though, if you want to beat a robot at chess, make it play on a real, physical chess board. " Goldberg adds.