Watch a robot trained by AI dribble, beat defenders, and score goals ( Video : Google DeepMind).
Humanoid robots have long been the product of scientists ' ambitions to create a device that can mimic all human activities, thereby serving us for various purposes.
However, the development of humanoid robots still faces many limitations, because every movement needs to be programmed. This requires scientists to synthesize enormous amounts of data.
However, modern robotics technology has received significant assistance from AI. This has created a major turning point in the development and practical application of robots.
The self-learning capability of AI means that robots will not need to move according to pre-programmed instructions, but will be fully capable of performing techniques without direct human control.
In a new study, scientists at Google DeepMind trained the robot Robotis OP3 to play soccer using a technique called "deep reinforcement learning".
This is a machine learning training technique that combines various AI training methods. In it, the robot uses an algorithmic network that functions like artificial neurons, and is structured similarly to the human brain.
Robots trained by AI have developed extremely difficult-to-program emergency movement behaviors, such as cornering their feet, spinning around to deceive opponents, and more.
In simulated matches, AI-trained robots moved 181% faster, turned 302% faster, kicked the ball 34% faster, and took 63% less time to recover from falls compared to robots not trained using this technique.
From the recorded video, it can be seen that the robot was able to perform complex maneuvers such as dribbling, blocking, beating defenders, scoring goals, etc., relatively skillfully and at high speed.
The researchers added that the robots have also developed extremely difficult-to-program emergency movement behaviors, such as rotating at the corners of their feet, spinning around to deceive opponents, and more.
The findings suggest that this AI training technique can be used to create simple, yet relatively safe, movements in humanoid robots in general. From this starting point, robots can be enhanced with more complex movements and applied more widely in practical situations.
The same technology applied has also enabled robots to brew coffee on their own after a training period of just 10 hours.
Source
Comment (0)