Integrity Monitoring and Fault Detection in Integrated Navigation Based on the Kalman Filter Innovation Sequence

Abstract

For GNSS/INS integrated navigation systems for automated driving, it is necessary not only to provide continuous positioning solutions, but also to detect faults in a timely manner and provide credible risk bounds in complex environments. The current literature tends to talk about innovation-based detection, fault detection algorithms and integrity metrics individually and has not yet elucidated the propagation of detection results to warning, isolation and protection-level maintenance. Considering the Kalman filter innovation sequence as the primary statistical strand, this paper critically surveys approaches like single-epoch chi-square testing, sequential probability ratio testing, robust filtering, and the structured residuals and covariance drift monitoring and examines their applications in the process of control of the alert limit, time to alert, protection level, and integrity risk in urban-canyon integrated GNSS/INS navigation applications. The findings demonstrate the innovation sequence is the most important statistical interface between the state estimation, fault detection, and integrity monitoring. Fault detection algorithms can be transformed into the integrity capability of automated-driving navigation systems only by combining the detection, isolation, warning, and protection-level updating into a single closed loop.