Compared with inner solid structures, the inner hollow structures of turbine blades show excellent mechanical properties and better physical properties, e.g., excellent heat dissipation and diversion effects, and light weight. Its complex inner structures cannot be detected accurately by traditional testing methods. Among all non-destructive testing (NDT) methods, ultrasonic testing method has certain advantages in measurement accuracy and efficiency for measuring wall thickness values. The accuracy of measurement results is provided by four criteria: precision, accuracy, stability, and efficiency. However, ultrasonic testing method still faces new challenges that are barely explored, e.g., inner structures have unique characteristics of complex geometrical features, the measurement process need to meet high-precision positioning and quickly detecting requirements. In this paper, in order to improve the detection accuracy and efficiency, an industrial robot is used to substitute the existing detection methods: once a new blade is clamped by terminal mechanism, a new calibration method adopted the six-point positioning scheme was provided to solve the problem of rapid and accurate updating tool coordinate system. The support vector machine (SVM) system was used to output regression results of the adjustment parameters based on the geometric features of distorted wave shapes. Two kinds of measurement paths were used to measure the wall thickness values for accurate measurement. The experimental results show the superiority and effectiveness of our method in terms of measuring stability, accuracy and testing efficiency.