Robust Filtering Techniques for RTK Positioning in Harsh Propagation Environments.

Area of Science:

• Geomatics Engineering
• Navigation Systems
• Signal Processing

Background:

• Global Navigation Satellite Systems (GNSSs) are crucial for intelligent transportation systems, including autonomous driving.
• Precise positioning is essential but challenged by harsh environments like urban areas, multipath, and signal disturbances.
• Conventional filtering techniques struggle with the complexity of resolving integer phase ambiguities under non-nominal conditions.

Purpose of the Study:

• To design robust filtering solutions for real-time kinematic (RTK) positioning.
• To address the mixed integer- and real-valued estimation problem in GNSS.
• To mitigate the impact of outliers and non-nominal measurement behaviors in challenging propagation conditions.

Main Methods:

• Exploration of Kalman filter (KF) families utilizing robust statistics and variational inference.
• Implementation of generalized M-based KF and variational-based KF approaches.
• Performance assessment using simulated scenarios and real-world measurement campaign data.

Main Results:

• Proposed robust filtering solutions demonstrate excellent resilience against outlying observations.
• The variational-based KF exhibited superior performance in terms of Gaussian efficiency and robustness.
• Effective mitigation of non-nominal measurement behaviors was achieved.

Conclusions:

• Robust filtering is vital for reliable GNSS-based precise positioning in adverse conditions.
• Variational inference provides a powerful framework for developing robust navigation filters.
• The variational-based KF presents a promising solution for challenging RTK applications.