With the continuous development of sensor technology, more and more intelligent sensors are applied to the detection field of industrial robots. In view of the problems of high cost, low accuracy, complex operation and light interruption that can not be achieved in the current measurement method, a method of measuring the performance of industrial robot using angle sensor measurement system is first proposed. A double angle sensors system model based on STM32 was built, the working principle and experimental method of the system were studied, and the error characteristics of the calibrated data were analyzed. The idea of least square method (LSM) was used to analyze the residuals characteristics of the system calibration. The method of iterative approximation of characteristic threshold was adopted to improve the accuracy and reliability of the measurement system, and the accuracy was evaluated as 0.15 mm. After many iterations, the drift error of the integral accumulation of the system was reduced, and the stability of the measurement system was improved. The experimental results show that the system is practical, low cost, stable and reliable, and has high precision, it provides a new method for the performance detection of industrial robots.