9634735 Kovadevic Automatic gas tungsten arc welding is widely used for precision joining. Although automatic welding machines can realize the pre-programmed welding parameters more accurately than a human operator, they lack the visual feedback capability associated with the human operator to compensate for the deviations in the process. This project is aimed to provide an innovative weld penetration control technology by developing a neurofuzzy model-based nonlinear adaptive control system with vision feedback of the weld pool. In the proposed control system, the welding current, torch speed, and arc length are used as the control variables to achieve the desired size and shape of the weld pool. Also investigated in the project are the dynamic behavior of the weld pool and the correlation between the pool geometrical appearance and penetration. Weld penetration control is central to automatic welding. The development of an advanced weld penetration control technology is essential to resolving the present bottle-neck problem in precise joining. The investigation of the dynamic behavior of the weld pool provides fundamental knowledge valuable to welding physics. Industrial relevancy of the research is enhanced by the existing strong collaboration with industrial firms. ***