These strange virtual creatures mutate their bodies to solve problems

“It’s already known that some bodies speed up learning,” Bongard said. “This work shows AI can search for those bodies.” Bongard’s lab makes robot bodies adapted for specific tasks, such as providing callus -like coatus on the feet to minimize wear and tear. Gupta and his colleagues came up with this idea, according to Bongard. “They show that the right body can also speed up changes in the robot’s brain.”

Ultimately, this approach could reverse the way we think about building physical robots, according to Gupta. Instead of starting with a fixed body configuration and then training the robot to perform a specific task, you can use DERL to allow the best body plan for that task to develop. and after that construction.

Gupta’s disagreements are about a vast shift in how researchers think about AI. Instead of teaching A’s specific tasks, such as playing Go or analyzing a medical scan, researchers are beginning to throw bots into virtual sandboxes-such as POET, The virtual arena to hide and seek in OpenAI, ug The virtual playground of DeepMind XLand—And get them to learn how to solve multiple tasks in constantly changing, open end training dojos. Instead of mastering a challenge, AIs are trained in this way to determine the overall skill.

For Gupta, exploring free form will be key for the next generation of AIs. “We need really open environments to make intelligent agents,” he said.

Source link


Leave a Reply

Your email address will not be published. Required fields are marked *