Design and experimental evaluation of block-pulse functions and Legendre polynomials observer for attitude-heading...

Area of Science:

• Navigation Systems Engineering
• Control Theory
• Sensor Fusion

Background:

• Micro electro mechanical systems (MEMS) sensors in Attitude and Heading Reference Systems (AHRS) suffer from time-varying errors like bias, scale factor instability, nonlinearity, and random walk.
• These errors degrade the accuracy and stability of low-cost MEMS-based navigation systems over time.
• Developing high-precision observers is crucial for improving the performance of MEMS AHRS.

Purpose of the Study:

• To design and implement a novel linear observer for MEMS-based AHRS.
• To address sensor and modeling uncertainties inherent in MEMS sensor data.
• To enhance the accuracy and reliability of attitude and heading estimation.

Main Methods:

• Utilized the duality principle between controllers and estimators in linear systems as the foundational design approach.
• Applied Legendre polynomials and block-pulse functions to solve linear time-varying control problems.
• Derived the Block-Pulse Functions and Legendre Polynomials Observer (BPLPO) from the control solution via duality theory.
• Simplified the optimal control problem into algebraic equations using hybrid function properties and operational matrices for efficient implementation.

Main Results:

• The proposed Block-Pulse Functions and Legendre Polynomials Observer (BPLPO) was successfully designed and implemented.
• The BPLPO effectively simplifies complex optimal control problems into manageable algebraic equations, suitable for low-cost systems.
• Vehicle field tests in urban environments demonstrated superior performance of the MEMS AHRS with BPLPO compared to the Extended Kalman Filter (EKF).

Conclusions:

• The BPLPO offers a robust and accurate solution for enhancing MEMS-based AHRS performance.
• The observer design effectively mitigates sensor and modeling uncertainties, leading to improved navigation accuracy.
• The proposed method provides a computationally efficient alternative to traditional filters like EKF for low-cost navigation applications.