A New Technique for INS/GNSS Attitude and Parameter Estimation Using Online Optimization
This addresses a specific engineering problem in navigation systems for vehicles, representing an incremental improvement over existing Kalman-like filtering methods.
The paper tackles the problem of attitude initialization failure in INS/GNSS integration by proposing an online constrained-optimization method that simultaneously estimates attitude, lever arm, and sensor biases without requiring prior information. Numerical results validate its effectiveness for high-accuracy applications.
Integration of inertial navigation system (INS) and global navigation satellite system (GNSS) is usually implemented in engineering applications by way of Kalman-like filtering. This form of INS/GNSS integration is prone to attitude initialization failure, especially when the host vehicle is moving freely. This paper proposes an online constrained-optimization method to simultaneously estimate the attitude and other related parameters including GNSS antenna's lever arm and inertial sensor biases. This new technique benefits from self-initialization in which no prior attitude or sensor measurement noise information is required. Numerical results are reported to validate its effectiveness and prospect in high accurate INS/GNSS applications.