Unsupervised robotic learning is a field within artificial intelligence, automation, and robotics. It describes a method where robots independently learn new tasks or behaviours without direct human instruction. This means robots try out on their own what works and what doesn't – similar to how children learn through trial and error.
In this way, the robot does not receive a fixed programme, but rather gathers its own experiences and draws conclusions from them. It remembers successful attempts and avoids mistakes in the future. This means the robot gets better and better at its tasks over time.
A vivid example: A cleaning robot is supposed to clean a flat autonomously but doesn't know its floor plan. Through unsupervised robotic learning, the robot drives through the rooms, bumps into furniture or walls, but remembers successful routes and obstacles. After a few runs, it knows how to clean all rooms efficiently and without bumping into anything – entirely without human help.
Unsupervised robotic learning is therefore an important step towards autonomous robots that can react flexibly and intelligently to their surroundings. This offers great advantages for industry as well as for everyday life.
Deutsch 
English (UK)