Self-Calibration of INS/Odometer Integrated Navigation System With a Large Initial Position Error

Inertial navigation system (INS) and odometer-integrated navigation system are widely used in land navigation. Prior calibration methods for INS/odometer systems require the aid of landmarks with precise positions or a global navigation satellite system (GNSS). GNSS cannot be used in some cases. When lots of INS/odometer systems are required to be simultaneously calibrated using the traditional calibration methods without GNSS aided, many landmarks must be prepared in advance. Additionally, some autonomous INS/odometer systems may need to be in-field calibrated without any landmark nearby and without GNSS aided. To reduce the calibration time and the requirement for initial position precision, a rapid self-calibration method under a large initial position error condition is proposed in this article. First, we propose a rapid analytic coarse calibration algorithm, which just needs a short straight path such as 10 m with a rough initial position. This algorithm removes the calibration error resulting from the inertial measurement unit (IMU) azimuth error. Then an odometer denoising algorithm is designed to improve the coarse calibration accuracy, especially when the trajectory is short. At last, we demonstrate the adaptability of the optimal estimation-based fine calibration algorithm to a large initial position error. Field test results show that the proposed self-calibration method achieves accurate calibration with an initial position error of 100 km class. And its calibration precision is close to that of the traditional method aided by landmarks with precise positions.