Google’s DeepMind AI Lab trained a robot to play soccer.
Jonathan Klotz Published
Google’s Bard AI may have had trouble learning when it was first launched, but another division, the DeepMind AI Lab, has made great strides in machine learning. Robot he might be able to join the NYPD. Robots are becoming more humanoid, but they are also learning to replace the best athletes on the pitch. As part of a Deep Reinforcement Learning (Deep RL) experiment, two pairs of adorable little machines learned to play soccer.
These robots do not operate according to specific parameters input by human engineers. Rather, we are adjusting the angle and position of the ball on the fly. Deep RL algorithms allow robots to operate without the need for constant inputs or pre-programmed conditions. This learning style is nothing new, and in 2013 DeepMind set up a program to play his Atari game. More recently, it was used to force robots to solve the Rubik’s Cube.
Part of the Deep RL programming also trained the robot to stand up, turn around, and run after a ball. In the video, you can see two cute robots flying around, but as you learn, you’ll feel more confident in what you do as the Learning Allogro Rim kicks in.
Despite one setback that even World Cup teams sometimes struggle with, the experiment was a huge success for the research team. By limiting themselves to open-source robot designs rather than building robots from scratch like Boston Dynamics, they realized that the results once limited to science fiction movies no longer require million-dollar hardware. proved not necessary.
Disney has gone in a different direction with its recent exhibit of emotionally intelligent robots. increase. It is when Deep RL is combined with emotional intelligence that it becomes truly dangerous and can lead to the singularity.
There are already robots playing soccer, and robots have their own tournament, the RoboCup. His two pints of Google a baller his size are unlike any tournament contestant. It wasn’t designed specifically for soccer competition, it had to be learned through trial and error and frequently rewarded for good behavior. Google has yet to explain how it achieved a positive reward for the robot, but if the comics have taught us anything, it’s likely that it was Sprocket.
Boosted by the launch of ChatGPT late last year, the AI space is growing so fast and so fast day by day that no one knows where it will go. It’s a cute soccer robot, but a machine that learns and adapts in real time as it picks up trash, drives a car, or wages war on the battlefield is something else entirely.
For now, Deep RL is still severely restricted from being placed at its most powerful range, so it looks safe from the robot revolution. However, Chat-GPT 3 has obvious limitations and the recent Chat-GPT 4 upgrade is significantly more powerful.
