Robot learning through observation and measurement in production process

The aim of the presented research is to to verify the hypothesis of the possibility of robot's learning, through observation and measurement, to perform production tasks. At present, the role of industrial robots in majority still comes down to precise re-enacting movements in accordance with changeless, strictly defined programme. The robots' adaptivity is the next step in the pursuit of automation of manufacturing operations, ensuring the high level of safety and productivity. The development of robotics towards the adaptability and machine intelligence requires the effort to automate the processes of the identification and analysis of actions, as well as drawing conclusions concerning their outcomes and the way of their performance. Such capacities are necessary for the autonomous operation of machines, consisting in shaping and applying operation patterns in particular circumstances. Assuming that, as a result of the research, the pattern of operations allowing the machine learning in the suggested way would be elaborated, the new paths of development in the industrial robotics would emerge. The article concerns the main concept of solutions and the plan of the research leading to the definition of the principles of the process of robot learning, through observation and measurement, to perform production tasks.