Robots in industrial manufacturing are usually programmed to repeatedly perform a fixed set of movements in an identical manner. However, rapidly changing product lines and the need for fast development and prototyping of new products combined with the expensive re-programming of such robots, demand a paradigm shift in industrial robotics. Robots in modern industry have to perform multiple tasks and must adapt to dynamic environments and change sin the executed task. In addition robots should be able to learn from human demonstrations in order to eliminate the costs of re-programming. Ideally the instructor should be allowed to teach the robot similarly to a novice co-worker. By teaching the robot personally, the robot adapts to the specific work flow of the worker and actively increases the productivity, thus the robot would act as a personalized assistant.