Machine learning by moving objects

In a new research project done at Bonn-Rhein-Sieg Universities in Europe, researchers aim to teach robots about their environment by moving objects and their interaction with them. The experience that the robot develops form the input received after doing certain moves improves the robots prediction about the future events at its surroundings. Similar to human learning, the robot uses its sensors to receive input and process them. The prediction of its own actions become more and more clear with each experience the robot develops, similar to human learning process.

Reference: www.physorg.com/news201880504.html