Applications of Learning Algorithms to Industrial Robotics

Learning algorithms are becoming popular in industrial manufacturing thanks to their promise to make a robot conscious of its surroundings and capable of human-like abilities, gaining greater flexibility with respect to traditional robotic systems. The aim is to operate also complex tasks without the need for explicit instructions, permitting the creation of fully autonomous systems where human operators are not included. A panoramic of the current state of the art in industrial fields is presented, starting from object recognition and grasping pose detection, to task planning and applications based on demonstrations by the operator.