The seam tracking sensor is indispensable for the robotic welding system to improve welding quality and productivity. Welding current has been widely used as the arc sensor in GMA welding. In this work, arc resistance and arc light signals are utilized. They have advantages over welding current : signal amplification effect for arc resistance and small fluctuation for arc light. Arc resistance is calculated from measured welding current and welding voltage. Arc light signal is detected using arc light sensor that consists of light collecting lens and optical fiber guide and photo diode. Signal processing methods such as quadratic curve-fitting and integration methods are compared through experiments. There is little difference in performance of both processing methods. It is also found that arc resistance and arc light show better tracking accuracy when using argon gas and $CO_2$ gas, respectively.