A New Self-Alignment Method without Solving Wahba Problem for SINS in Autonomous Vehicles

Overview

• This paper presents a new self-alignment method for strapdown inertial navigation systems (SINS) under swaying conditions.
• The proposed method can determine the latitude and attitude simultaneously without relying on external aids or solving the Wahba problem.
• The method constructs a dyadic tensor from observation and reference vectors, which allows all equations to be accumulated into a single equation.
• This enables the latitude variable to be extracted and solved based on the same eigenvalues, while the attitude is obtained through eigenvalue decomposition.
• Simulation and experimental tests demonstrate the effectiveness of the proposed method, with improved convergence speed and stability compared to the TRIAD method, and comparable alignment accuracy to the OBA method with or without latitude information.

Plain English Explanation

Self-alignment in inertial navigation systems is crucial for providing accurate initial state information, such as the vehicle's attitude and position, without relying on external aids. This is particularly important in situations where the position is not needed or available, like in attitude heading reference systems.

The paper introduces a new self-alignment method that can determine the latitude and attitude simultaneously, even when the vehicle is swaying or moving unpredictably. Instead of solving the complex Wahba problem, the researchers use a more straightforward approach. They construct a special mathematical object called a "dyadic tensor" from the observation and reference vectors, which allows them to combine all the relevant equations into a single equation.

From this single equation, they can extract the latitude variable and solve for it using the same eigenvalues on both sides of the equation. The attitude is then obtained through a standard eigenvalue decomposition process. This novel approach eliminates the need to solve the Wahba problem, which is typically used in inertial-visual alignment methods.

The researchers' simulations and experiments show that their method outperforms the TRIAD algorithm in terms of convergence speed and stability, and matches the accuracy of the OBA method, even when latitude information is not available. This makes the proposed self-alignment technique particularly useful for autonomous vehicle applications where inertial measurement units and GPS are used for navigation and localization.

Technical Explanation

The paper presents a new self-alignment method for strapdown inertial navigation systems (SINS) that can determine the latitude and attitude simultaneously under swaying conditions. Unlike existing methods that rely on solving the Wahba problem, the proposed approach constructs a dyadic tensor from the observation and reference vectors, allowing all equations related to these vectors to be accumulated into a single equation.

By extracting the latitude variable from this single equation based on the same eigenvalues on both sides, and then obtaining the attitude through eigenvalue decomposition, the method avoids the need to solve the Wahba problem. Simulation and experimental tests demonstrate that the proposed self-alignment method outperforms the TRIAD algorithm in terms of convergence speed and stability, and achieves comparable alignment accuracy to the OBA method, with or without latitude information.

The key technical contributions of the paper include:

1. Constructing a dyadic tensor from observation and reference vectors to consolidate all relevant equations into a single equation.
2. Extracting the latitude variable from the single equation using the same eigenvalues on both sides, without solving the Wahba problem.
3. Obtaining the attitude through eigenvalue decomposition of the consolidated equation.
4. Validating the effectiveness of the proposed self-alignment method through simulation and experimental tests, showing improved performance compared to TRIAD and comparable accuracy to OBA.

Critical Analysis

The paper presents a novel and promising self-alignment method for SINS that addresses some of the limitations of existing approaches. The key strength of the proposed method is its ability to determine both latitude and attitude simultaneously without relying on external aids or solving the Wahba problem, which can be computationally intensive.

However, the paper does not provide a thorough analysis of the method's robustness to various environmental conditions or sensor noise levels. Additionally, the experimental validation is limited, and the authors do not discuss the potential challenges or limitations of the method in real-world applications.

Further research could explore the method's performance under more diverse and challenging scenarios, such as high-dynamic motions, sensor failures, or GPS outages. Comparisons with other self-alignment techniques, such as QUEST or EKF-based methods, could also help provide a more comprehensive evaluation of the proposed approach.

Overall, the paper presents an intriguing self-alignment method that could be valuable for autonomous vehicle applications, particularly in scenarios where position information is not readily available. Further research and real-world validation would be needed to fully assess the method's capabilities and practical implications.

Conclusion

This paper introduces a new self-alignment method for strapdown inertial navigation systems that can determine latitude and attitude simultaneously under swaying conditions. By constructing a dyadic tensor from observation and reference vectors, the method avoids the need to solve the computationally intensive Wahba problem, which is typically used in inertial-visual alignment approaches.

The proposed approach demonstrates improved convergence speed and stability compared to the TRIAD algorithm, while achieving comparable alignment accuracy to the OBA method, even in the absence of latitude information. This makes the method particularly useful for autonomous vehicle applications that rely on inertial measurement units and GPS for navigation and localization.

While the paper provides promising initial results, further research is needed to fully assess the method's robustness and applicability in diverse real-world scenarios. Nonetheless, the novel approach presented in this work represents a valuable contribution to the field of inertial navigation system alignment and could have important implications for the development of autonomous vehicle technologies.