GNSS-Denied Failure Simulator

What you are looking at

A planar UAV flies a mission with an IMU-driven EKF aided by GNSS position (innovation-gated and cross-checked against odometry), camera velocity and LiDAR velocity. Health monitors feed a navigation confidence level: GREEN (GNSS and odometry available), AMBER (degraded but bounded), RED (no velocity aiding or large uncertainty). A mode manager chooses NOMINAL, GNSS-ONLY, GNSS-DENIED, DEAD RECKONING, or FAILSAFE (a controlled stop).

Tick faults and set their start and duration, or use a scenario button. The timeline shows each sensor’s availability, the confidence level and the mode; the event log records what the monitors noticed and when.

Try this

1. Run “GNSS loss” with the camera and LiDAR fitted, then untick both: compare error growth and the time to failsafe.
2. Run “Spoofing”, then “Spoofing, checks off”. Slow the ramp to 0.2 m/s: when does the innovation gate stop catching it and the cross-check become the only defence?
3. Remove the camera and LiDAR (untick “fitted”) and spoof GNSS. Without aiding sources there is nothing to cross-check against: what happens?
4. Inject an IMU bias step during GNSS loss and watch the estimator’s covariance and error. Which monitor, if any, notices?
5. Set “Failsafe after RED for” to 30 s versus 5 s with all aiding sensors lost: what is the trade-off between nuisance failsafes and drift?

Model and its limits

Included: a 9-state planar EKF with GNSS position, camera velocity (with a scale state) and LiDAR velocity updates, a 99% NIS innovation gate on GNSS, a windowed GNSS-versus-odometry cross-check, feature-count and data-timeout health checks, a confidence classifier, a mode manager with failsafe timers, and a time-warped controlled stop.

Not included: a real flight controller, 3-D attitude, specific sensor products, jamming at the RF level (it appears as simple loss), or certified integrity monitoring. The thresholds are teaching defaults, not recommended values: choosing and justifying them for a real vehicle is the subject of module 8 and the final capstone.

Related lessons

• M2-T5: GNSS signal blockage
• M2-T6: GNSS jamming and spoofing
• M2-T16: Multi-sensor GNSS-denied navigation architectures
• M3-T13: Innovation monitoring
• M3-T14: Normalized Innovation Squared
• M3-T16: Sensor fault detection and isolation
• M8-T9: Fault detection, isolation and recovery
• M8-T10: Degraded navigation modes
• M8-T11: GNSS loss handling
• M8-T12: Visual odometry failure handling
• M8-T13: IMU failure handling
• M8-T14: Communication loss and failsafes