Hi, this is Wayne again with a topic “Step By Step: How Google DeepMind AI Is Learning To Walk ?”.
Ai can learn how to walk. In the recent years, AI technology has been used by Google for experimentation, utilizing a project known as deepmind. This experiment is based on the use of RL reinforcement, learning to make a computer navigate, complicated and novel environments. Now, an AI that is capable of learning to walk by itself has been introduced by the tech giant, the outcomes of which might be hilarious and scary.
Three prototypes have been deployed based upon the number of legs. These prototypes have sensors placed in them which provide them. The information about the location to go from points A to B, but this was not enough to provide them the ability to walk for these models.
It was almost like Reinventing walking, but their gate is quite similar to human beings, except that they exhibit a bit more of arm flailing than humans. Although the robot took a couple of Falls from time to time, it succeeded mostly in maintaining its balance and jumping over some hindrances. It did often sway away from its training track and fall down too, but hey.
We can’t expect to learn to walk without falling when the Prototype Aces its ongoing test. The automated generator makes another test for the Prototype. Every new challenge produced is based on the robot’s training memory and assists in spreading its abilities over a huge scope of new tasks.
Thank you. .