Develop an FMEA for a GNSS-denied UAV

What you will do

You will write an FMEA for the navigation subsystem of the Capstone 4 vehicle, using a data file so that it can be checked automatically.

Before you start

• Lesson M8-T7 (FMEA) and the architecture of lesson M2-T16.
• The Architecture Builder and Failure Simulator for exploring failure behaviour.

Steps

1. List the functions of the navigation subsystem from your architecture diagram (at least: GNSS positioning, visual odometry, LiDAR odometry, IMU propagation, fusion, monitoring, mode management, PX4 interface).
2. For each function list at least two failure modes (loss, degradation, wrong output). Include wrong but plausible outputs, which are the dangerous ones.
3. Describe local effect, system effect and vehicle-level effect. Score severity, occurrence and detection with scales you define in a table at the top of the document.
4. Record the existing control or monitor for each mode and a time-to-detect estimate.
5. Add recommended actions, an owner and a verification (test ID or simulation scenario).
6. Compute RPN, mark severity ≥ 9 modes, and check that every marked mode has a verified mitigation, using the script from lesson M8-T7.

Use the failure simulator as evidence

• Inject each GNSS, camera, LiDAR and IMU failure and note the detection time from the event log.
• Use those times in the detection column and as evidence for the detection score.
• Find a failure combination the architecture does not handle and add it as a new row with an action.

How to know it worked

• At least 20 failure modes across at least 6 functions with scoring scales defined.
• Every severity ≥ 9 mode has an action and a verification ID.
• At least five detection claims are supported by simulation evidence.

If it does not work

References

• SAE ARP4761
• NASA Systems Engineering Handbook

Record your progress