Robust Attitude Estimation for Low-Dynamic Vehicles Based on MEMS-IMU and External Acceleration Compensation

Area of Science:

• Robotics and Control Systems
• Sensor Fusion and Navigation
• Mechanical Engineering

Background:

• Micro-electro-mechanical system inertial measurement units (MEMS-IMUs) are crucial for attitude determination.
• Non-gravitational accelerations significantly impact MEMS-IMU attitude estimation accuracy.
• Existing methods struggle with low-dynamic vehicles under vibrations and uneven terrain.

Purpose of the Study:

• To develop a robust Kalman filter (RKF) for improved MEMS-IMU attitude estimation.
• To enhance the accuracy and reliability of attitude determination for low-dynamic vehicles.
• To compensate for external acceleration disturbances in real-world conditions.

Main Methods:

• Constructed a state model for MEMS-IMU attitude estimation using a simplified direction cosine matrix.
• Developed an RKF that adaptively adjusts noise covariance to compensate for external acceleration.
• Estimated unmodeled external acceleration variance online using filtering innovations.
• Validated the RKF using experiments on a three-axis turntable, an automatic vehicle, and a tractor tillage test.

Main Results:

• RKF achieved a root mean square error (RMSE) of 0.051° on a turntable, outperforming conventional Kalman Filter (KF) and MTi-300.
• Dynamic attitude estimation for an automatic vehicle showed smoother pitch angles with RKF compared to KF, reducing RMSE from 0.875° to 0.460°.
• Tractor tillage tests demonstrated a 47.5% improvement in plough pitch RMSE with RKF (0.259°) compared to KF (0.493°).

Conclusions:

• The developed RKF effectively suppresses external acceleration disturbances in MEMS-IMU-based attitude estimation.
• RKF offers superior performance and accuracy for attitude determination in challenging vehicle dynamics and environments.
• The adaptive noise covariance compensation is key to the RKF's enhanced robustness and reliability.