Posted 4/11/2024, 2:20:20 PM
Soccer-Playing Mini Humanoid Robots Learn Complex Skills Through AI Training
- Scientists trained miniature humanoid robots to play soccer using a new AI technique called "deep reinforcement learning"
- The AI training allowed the robots to move faster, turn quicker, kick better, and recover from falls compared to untrained robots
- The robots developed new behaviors like pivoting and spinning that are hard to directly program
- The training method allowed the robots to learn complex movements and strategies through experience
- This AI training technique could lead to more sophisticated robot motions and abilities to handle complex situations