Research on GNSS/MEMS IMU Array Fusion Localization Method Based on Improved Grey Prediction Model

Area of Science:

• Robotics
• Navigation Systems
• Signal Processing

Background:

• Global Navigation Satellite System (GNSS) signals are prone to interference and blockage in vehicle navigation, degrading positioning accuracy.
• Traditional grey prediction models struggle with the complex motion data characteristic of vehicles.
• Accurate real-time positioning is critical for autonomous and assisted driving systems.

Purpose of the Study:

• To develop an enhanced GNSS/MEMS IMU fusion localization method for vehicles.
• To improve positioning accuracy and reliability, particularly under GNSS signal denial conditions.
• To overcome the limitations of existing grey prediction models in dynamic environments.

Main Methods:

• A multi-feature fusion GNSS confidence evaluation algorithm assesses GNSS data reliability in real-time.
• An improved grey prediction model incorporates dynamic background value optimization and residual sequence compensation for complex motion data.
• An adaptive fusion framework integrates GNSS and MEMS IMU data, using grey model predictions as virtual measurements during GNSS outages.

Main Results:

• The improved grey prediction model achieved 31%, 52%, and 45% higher accuracy than the traditional GM(1,1) model in straight, turning, and acceleration scenarios, respectively.
• Positioning accuracy improved by over 79% compared to pure Inertial Navigation System (INS) methods during a 30-second GNSS denial period.
• The proposed method demonstrated enhanced sensitivity to vehicle motion state changes and improved nonlinear motion prediction.

Conclusions:

• The proposed GNSS/MEMS IMU fusion method significantly enhances vehicle positioning accuracy and robustness.
• The improved grey prediction model effectively handles complex vehicle dynamics and GNSS signal disruptions.
• This approach offers a reliable solution for navigation systems facing intermittent or denied GNSS availability.