Abstract: With the continuous expansion of global energy pipeline networks, high-precision positioning technology for in-pipe inspection robots has become critical for ensuring operational safety and efficiency. In enclosed pipeline environments where satellite signals are absent, conventional inertial navigation system (INS)/odometer(ODO) integrated positioning methods exhibit significant error accumulation, failing to meet the requirement for accurate long-distance defect localization. To address this issue, this study proposes an IMU/ODO state recognition and constrained positioning algorithm designed for pipeline scenarios, aiming to systematically resolve the organic integration of feature recognition and motion constraints in such settings. By fusing data from the IMU and odometer, a motion state recognition algorithm based on acceleration magnitude and gyroscope standard deviation analysis is developed, enabling accurate discrimination of stationary and straight-line motion states. Leveraging the geometric constraint characteristics of pipelines, constraint conditions such as zero-velocity constraints and attitude constraints are introduced according to different motion states to suppress the divergence of INS errors. Experimental results demonstrate that in a 64 m simulated straight pipeline test, the proposed method reduces the eastward root mean square error (RMSE) by 70.5%, the northward RMSE by 55.3%, and the vertical RMSE by 40.0%. In a 64 m simulated curved pipeline test, the eastward RMSE decreases by 49.1%, the northward RMSE by 57.1%, and the vertical RMSE by 30.0%. In a 96 m field pipeline experiment, the terminal positioning accuracy is better than 0.1 m, significantly outperforming the traditional INS/ODO integration method. The proposed approach operates without external auxiliary information and provides a high-precision positioning solution for GNSS-denied pipeline environments.