Session A3: Aviation, Aeronautics, and Uncrewed Aerial Applications 1

Kalman Filter Integrity Monitoring Over Finite Time-Intervals
Matthew Marti and Mathieu Joerger, Virginia Tech
Location: Holiday 6 (Second Floor)
Date/Time: Thursday, Sep. 11, 11:03 a.m.

In this paper, we derive and analyze a Kalman filter (KF) integrity monitoring method for measurement faults occurring over a finite past-time interval. The method is intended for safety-critical navigation applications involving GNSS corrections with latency or delayed GNSS authentication, i.e., where a delayed past-time integrity guarantee is available and a current-time integrity risk bound is needed. In-between these two times, redundancy of sensor information is assumed even if all recent GNSS measurements are simultaneously faulted or spoofed, for example, using inertial measurement units (IMUs). To ensure navigation integrity, we derive a compact and recursively updated fault detector using KF innovation projections instead of the conventional normalized innovations squared (NIS) or accumulated NIS. The integrity monitor can then be tailored to the estimated state of interest, thereby providing tight integrity risk bounds while accounting for worst-case fault profiles. We evaluate this integrity risk bound in an example tight integration of GNSS and IMUs.