Detection of Sensor Faults in Small Helicopter UAVs Using Observer/Kalman Filter Identification

Table 5

Comparison of results: sensor fault detection with drift of 25% of maximum sensor value for OKID, ARX, correctly tuned Kalman filter, and a Kalman filter with 20% error parameter estimation.

Mean detection time		GPS			ACCEL			MAG			GYRO
	𝑋	𝑌	𝑍	𝑋	𝑌	𝑍	𝑋	𝑌	𝑍	𝑋	𝑌	𝑍
OKID	0.13	2.31	0.22	0.42	1.33	0.45	0.29	0.81	0.32	0.42	1.08	0.19
ARX	0.15	2.25	0.28	0.50	1.36	0.40	0.18	0.89	0.40	0.39	1.18	0.19
Kalman	0.13	2.40	0.30	0.51	1.29	0.39	0.29	0.85	0.38	0.45	1.09	0.27
Kalman, 20% error	0.25	3.89	0.52	1.32	—	2.95	1.47	0.97	0.52	0.72	—	1.31