Machine learning has a tight bond with Artificial intelligence. It is a branch of computer science implemented in several ways and methods. Machine learning deals with those mechanisms that support intelligent machines to improve their performance and skills. Robotic systems, thanks to this technology, will be able to refine their abilities and, over time, they will learn to take decisions in accordance with previous acts.
There are different learning techniques, different not only for the algorithm implemented, but especially they differ for the purpose. These ways are divided into three systems:
- supervised (the machine is given a database including knowledge and experiences and it will use them when facing a problem)
- not supervised (information given to the robot is not coded, therefore the it does not know when to use it. Therefore, the machine can decide which response to use and, thanks to experience, it will understand if it is the best one)
- reinforcement (this is the most complex machine learning system, as the robotic system needs to be equipped with tools that facilitate its learning and understanding of the surrounding environment)
Machine learning ensures that machines get deeper and deeper learning to enable them to enrich the knowledge and to understand when to implement it. Improving these technologies will help robots reacting to the environmental situations in a more efficient way.