At one point as a little kid, you learned how to catch yourself after a fall, and eventually you learned how to walk on your feet. You likely got encouragement from your parents, but you learned mostly through trial and error. This is not what robots look like spot And the atlas From Boston Dynamics, you know that Walking and dancing. They are meticulously coded to handle the tasks we throw at them. The results can be impressive, but it can also make them unable to adapt to situations that their programs do not cover. A joint team of researchers from Zhejiang University and the University of Edinburgh claim to have developed a better method.

at The last paper Published in the journal Robotics scienceThey detailed the AI ​​augmentation approach they used to allow their dog-like robot, Jueying, to learn how to walk and recover from falls on its own. The team said Wired First they trained a program that could direct a virtual version of the robot. It consists of eight “experts” in artificial intelligence who are trained to master a specific skill. For example, one became proficient in walking, while the other learned how to balance. Every time the digital robot successfully accomplishes a task, the team rewards it with a virtual point. If all this sounds familiar, it’s because it’s the same approach Google has recently used to train its innovations. The MuZero algorithm