Vehicle navigation system

1. A method for controlling a vehicle, the method comprising:

receiving a plurality of measurements from a plurality of sensors comprising a set of non-inertial sensors;

determining a first group mean from the measurements;

determining a vehicle state prediction based on a previous vehicle state and a motion model coupling each of the plurality of measurements, wherein the vehicle state prediction comprises a prediction value for each of the plurality of measurements;

determining measurement faults within the plurality of measurements based on the prediction value for each of the plurality of measurements, wherein determining the measurement faults within the plurality of measurements includes:

identifying a sensor of the plurality of sensors having an output that is furthest from the first group mean;

calculating a second group mean without the output from the identified sensor; and

comparing the second group mean against a rolling average;

generating an updated set of measurements from the plurality of measurements based on the measurement faults; and

determining an updated vehicle state based on the updated set of measurements, the vehicle state prediction, and an observation model.

2. The method of claim 1 , wherein the motion model uses a subset of sigma-points computed from only one side of a distribution relative to an operating point.

3. The method of claim 1 , further comprising:

for each of the plurality of sensors, determining a measurement covariance that is decoupled from a vehicle state covariance update; and

modifying the observation model based on the measurement covariance for each of the plurality of sensors.

4. The method of claim 3 , wherein the measurement covariance is determined independently for a first set of the plurality of sensors, a second set of the plurality of sensors; and the non-inertial sensors.

5. The method of claim 4 , wherein the first and second sets of the plurality of sensors comprise distinct types of inertial sensors.

6. The method of claim 1 , wherein the plurality of sensors further comprises a set of inertial sensors.

7. The method of claim 1 , wherein the non-inertial sensors comprise a time of flight sensor and a GPS sensor.

8. The method of claim 7 , wherein the time of flight sensor comprises radar.

9. The method of claim 1 , wherein the non-inertial sensors comprise air data sensors.

10. The method of claim 9 , further comprising controlling an electric propeller of the vehicle based on the updated vehicle state, wherein the air data sensors comprise a propeller model associated with the electric propeller.

11. The method of claim 1 , wherein the determining measurement faults within the plurality of measurements comprises comparing measurements from dissimilar sensors using a plurality of comparison types.

12. The method of claim 11 , wherein the plurality of measurements comprise:

a first and second measurement, each associated with a first sensor type;

a third and fourth measurement, each associated with a second sensor type, different from the first sensor type;

wherein the first and third measurements are associated with a first reference frame, wherein the second and fourth measurements are associated with a second reference frame which is different from the first reference frame, and

wherein the plurality of comparison types comprises an individual fault test comparing the first measurement to a voted average of at least the second and fourth measurements.

13. The method of claim 1 , wherein each of the plurality of measurements is received as a plurality of redundant measurement signals, the method further comprising:

for each measurement of the plurality, determining the measurement from the plurality of redundant measurement signals using a first voting scheme;

determining a voted vehicle state from the updated vehicle state using a second voting scheme, and

controlling the vehicle based on the voted vehicle state.

14. The method of claim 13 , wherein the second voting scheme determines the voted vehicle state based on a Bhattacharyya distance of the updated vehicle state.

15. The method of claim 13 , wherein the updated vehicle state is determined by a first processor, wherein the voted vehicle state is determined from the updated vehicle state, a second vehicle state, and a third vehicle state, the second and third vehicle states independently determined by a second and third processor, respectively, based on the previous vehicle state.

16. A method for controlling a vehicle, the method comprising:

receiving a plurality of measurements from a plurality of sensors comprising a set of non-inertial sensors;

determining a vehicle state prediction based on a previous vehicle state and a motion model coupling each of the plurality of measurements, wherein the vehicle state prediction comprises a prediction value for each of the plurality of measurements;

determining measurement faults within the plurality of measurements based on the prediction value for each of the plurality of measurements, wherein the determining measurement faults within the plurality of measurements comprises comparing measurements from dissimilar sensors using a plurality of comparison types;

computing a measurement health for each measurement of the plurality of measurements based on the measurement comparisons;

generating an updated set of measurements from the plurality of measurements based on the measurement faults, wherein generating the updated set of measurements from the plurality of measurements comprises weighting the updated set of measurements according to the measurement health; and

determining an updated vehicle state based on the updated set of measurements, the vehicle state prediction, and an observation model.

17. The method of claim 16 , wherein a comparison type of the plurality comprises a rolling mean error computed over a window of previous measurements.

18. The method of claim 16 , further comprising: determining that a number of measurement faults exceeds a threshold number of faults; and including, within the updated set of measurements, a subset of the measurement faults based on the measurement health and the threshold number of faults.

19. The method of claim 16 , wherein the measurement health is computed based on a Mahalanobis distance associated with the measurement.

20. A method for controlling a vehicle, the method comprising:

receiving a plurality of measurements from a plurality of sensors comprising a set of non-inertial sensors;

determining a vehicle state prediction based on a previous vehicle state and a motion model coupling each of the plurality of measurements, wherein the vehicle state prediction comprises a prediction value for each of the plurality of measurements;

determining measurement faults within the plurality of measurements based on the prediction value for each of the plurality of measurements, wherein the determining measurement faults within the plurality of measurements comprises comparing measurements from dissimilar sensors using a plurality of comparison types;

computing a measurement health for each measurement of e plurality of measurements based on the measurement comparisons;

generating an updated set of measurements from the plurality of measurements based on the measurement faults; and

determining an updated vehicle state based on the updated set of measurements, the vehicle state prediction, and an observation model,

wherein the plurality of measurements comprise:

a first and second measurement, each associated with a first sensor type, and

a third and fourth measurement, each associated with a second sensor type, different from the first sensor type, the first and third measurements associated with a first reference frame, the second and fourth measurements associated with a second reference frame which is different from the first reference frame,

wherein the plurality of comparison types comprises:

a first common mode fault test comparing a voted average of at least the first and second measurements to a voted average of at least the third and fourth measurements, and

a second common mode fault test comparing a voted average of at least the first and second measurements to a voted average of at least the first, second, third and fourth measurements.

21. The method of claim 20 , wherein the plurality of comparison types further comprises a common mode fault tiebreak test, comprising comparing a voted average of at least the first and second measurements to the prediction value in a common frame.