Multi-sensor workflow analysis for assembly tasks

Computerized and virtual reality training systems - many of which utilize haptic feedback - have gained increasing acceptance and sophistication in recent years. Those tools such as laparoscopic surgical simulation, help to invent techniques not available with traditional training methods. In the past years not only the idea of training the machine was developed, but also of a machine acting as a trainer. In the fields of medicine, sport and entertainment, scientists are working on the development of systems, which can be used to teach human. Developing a machine which can perform human skill, poses an enormou challenge. After collecting the sensor data which characterizes the skill, this data should be transformed into a machine understandable form.This diploma thesis is concentrated on giving a solution to the challenging task of learning and recognition of skills. The complexity of this problem is already discussed in the previous two sections. The goal of the thesis is to introduce and discuss possible solution for reducing the complexity of the representation of a skill and its transfer to a machine.