The goal of this project is to develop the formalized knowledge of skill in terms of impendance parameters of performers and to use this knowledge for developing training and assistance systems with an interactive robot. This project will measure and make a comparative analysis of impedance parameters of skilled and unskilled manual welders. Pattern analysis will be performed on the impedance profiles to identify the constitutive patterns of skilled and unskilled impedance profiles. The patterns in both groups will be compared. The comparative analysis is important to understand what is to be reinforced and what is to be suppressed for a good performance. The knowledge of impedance characteristics of skilled and unskilled welders will be used to develop training and assistance systems with an interactive robot. The training system will detect the unskilled type movements in real-time and provide notice-feedback to the welder at the instant of the movement. This will let the welder be immediately aware of the causes of the errors. The training system will also provide off-line verbal instructions based on the performance of the welder. These instructions will be in the form such as “you go too fast”, “you hold it too loose”, and “you vibrate too much”. The assistance system will detect the arm impedance parameters of the performer in real-time. The system will have the knowledge of the impedance patterns of skilled performers. The assistive robot will compensate the impedance of the unskilled performer in the direction of that of an average skilled performer. The system will also estimate the dynamic skill-level. The level of impedance compensation will be adaptive, inversely proportion to the skill level.