Abstract: In the process of tunneling in underground coal mines, high dust, low illumination, and complex background conditions make it difficult to detect the attitude of the tunneling equipment, which has seriously affected the automation and intelligence process of tunnel tunneling. Therefore, a method for measuring the position and posture of cantilever roadheader based on binocular vision was proposed. Boom roadheader is currently the main equipment for roadway excavation, and its motion characteristics show slow movement and large swing. and the status quo of position and posture measurement technology of machine body of tunneling equipment were analyzed. Combined with the working characteristics of the roadheader, the binocular vision measurement scheme and measurement structure were constructed. Under the complex working conditions of low illumination, high dust and high water mist in coal mines, there was still serious lack of characteristic information. The infrared LED light source has good stability and penetrability and can well adapt to the complex driving environment. The Gaussian fitting method was used to realize the fast and accurate extraction of the characteristics of the spot center point. On this basis, a feature point ring matching method based on the time domain image sequence was proposed and its effectiveness was verified. The 3D-3D motion estimation method was used to achieve the pose parameters. In view of the limited distance of binocular vision measurement, the dual-target moving measurement strategy was designed to realize continuous and uninterrupted measurement.Finally, the experimental verification platform was built. The results show that the measurement error of the method in the X direction that isthe lateral displacement is ±30 mm,and the measurement error of heading angle is within ±1°, which meets the requirements of underground roadway excavation in coal mines.