| dc.description.abstract | The Global Navigation Satellite System (GNSS) is widely used in modern positioning and navigation applications. However, in obstructed environments, GNSS signals may be blocked or degraded, resulting in increased positioning errors. To enhance the accuracy of navigation systems under constrained or GNSS-denied environments—particularly in vertical height measurement and motion trajectory estimation—this study aims to integrate Micro-Electro-Mechanical Systems (MEMS) Inertial Measurement Units (IMUs) with a barometer to develop a fused navigation system, with its performance evaluated and compared against GNSS data.
The core methodology involves fusing sensor data from the IMU (accelerometer, gyroscope, and magnetometer) and the barometer. The Madgwick algorithm is used for attitude estimation, utilizing accelerometer and gyroscope data for trajectory estimation, while the magnetometer corrects heading drift to improve short-term relative positioning accuracy. In addition, pressure changes measured by the barometer are used for altitude compensation, and an Error-State Kalman Filter (ESKF) is applied to fuse the altitude information, enabling a stable navigation solution.
Due to the drift and integration errors of traditional IMUs, which accumulate over time and degrade long-term accuracy, this study focuses on using barometric data to reduce height estimation errors and improve short-term vertical accuracy. To address the instability of GNSS altitude data, we analyze the performance of GNSS and barometric data under various test environments. Multiple test scenarios were designed, including static tests and vehicle motion tests, to evaluate system performance under different dynamic conditions. For the vehicle motion test, OpenCV image processing techniques were introduced to extract dashboard footage from video recordings and recognize vehicle speed information, further improving speed estimation accuracy and mitigating the velocity drift typically associated with pure IMU data.
Experimental results show that in height estimation, GNSS altitude data yielded a root mean square error (RMSE) of up to 22 meters, while barometric altitude data achieved an RMSE of only 3 meters, highlighting the superiority of barometers in height compensation. Furthermore, trajectory error analysis for four test cases yielded PN-RMSE values of 1.93%, 3.48%, 2.53% and 2.69%, and DPE values of 0.005%, 0.42%, 1.09% and 0.16%, respectively. These results confirm that PN-RMSE remains below 3.48% and DPE under 1.09%, demonstrating the feasibility and effectiveness of the proposed method.
In summary, the developed system exhibits preliminary positioning capabilities. The fusion of IMU and barometer data significantly enhances vertical measurement accuracy, while the use of OpenCV for vehicle speed recognition improves speed estimation and overall positioning performance. This system can be effectively applied in GNSS-denied or constrained environments, such as in unmanned aerial vehicles (UAVs), autonomous vehicles, and indoor or underground navigation. Future work may include real-time implementation, environmental temperature compensation, machine learning algorithms, and higher-precision IMU sensors to further enhance the overall accuracy of the navigation system, providing a more reliable solution for precise navigation. | en_US |