An Acceleration Denoising Method Based on an Adaptive Kalman Filter for Trajectory in Merging Zones

Table 2

Comparison of different approaches using NGSIM I-80 data.

Variable		Raw data	Kalman filter	Simple moving average	Proposed method
Jerk	Std	33.40	7.98	2.95	3.41
	Mean	0.00	0.00	0.00	0.00
	Max	1640	91	66	53.67
	Min	−1846	−64	−59	−54.33
Jerk values higher than ± 15 (%)	Std	11.22	2.88	0.97	0.50
	Mean	22.83	0.30	0.29	0.16
	Max	69.23	22.16	11.35	4.17
	Min	4.70	0.00	0.00	0.00
1-s window with more than one sign inversion of jerk (%)	Std	10.31	12.64	14.16	4.95
	Mean	75.23	21.22	56.10	13.54
	Max	100	82.05	81.82	80.42
	Min	42.57	10	6.53	7.40