OP-SLAM: An RGB-D SLAMMOT Method Leveraging the Constraints of Object Planar Features | Wuhan University Journal of Natural Sciences

Open Access

Table 3

Object pose estimation results compared with other visual SLAMMOT methods on the KITTI tracking dataset

Method		seq-obj.id
Method		03-1	05-31	10-0	11-0	11-35	18-2	18-3	19-63	19-72	20-0	20-12	20-122
DynaSLAM^[16]	ATE	0.69	0.51	0.95	1.05	1.25	1.10	1.13	0.86	0.99	0.56	1.18	0.87
	$R P E_{t}$	0.34	0.26	0.40	0.43	0.89	0.30	0.55	1.45	1.12	0.45	0.40	0.72
	$R P E_{R}$	1.84	13.50	2.84	12.51	16.64	9.27	20.05	48.80	3.36	1.30	6.19	5.75
	2D-TP	50.00	28.96	81.63	72.65	53.17	86.36	53.33	35.26	29.11	63.68	42.77	34.90
TwistSLAM^[17]	ATE	0.31	0.35	0.77	0.17	0.10	0.21	0.15	0.28	0.16	0.17	0.24	0.17
	$R P E_{t}$	0.10	0.19	0.21	0.23	0.03	0.27	0.21	2.17	0.05	0.20	0.20	0.02
	$R P E_{R}$	0.28	0.58	1.98	0.23	0.11	0.66	0.56	1.08	0.34	0.72	1.54	0.07
	2D-TP	58.02	30.84	7.20	29.61	65.00	86.47	28.19	65.93	16.92	84.75	14.24	84.94
Yang et al^[19]	ATE	0.18	0.28	0.20	0.22	0.87	0.20	0.30	0.95	0.78	0.22	0.67	1.07
	$R P E_{t}$	0.20	0.16	0.15	0.25	0.71	0.10	0.33	1.01	0.32	0.17	0.15	0.99
	$R P E_{R}$	0.04	0.01	1.03	0.02	0.03	0.04	0.64	0.88	0.33	0.02	0.33	0.01
	2D-TP	90.90	100.00	100.00	100.00	85.96	90.23	95.33	100.00	95.56	88.96	100.00	97.59
DynaMeshSLAM^[8]	ATE	0.28	0.16	0.59	0.54	0.12	0.30	0.13	0.33	0.28	2.58	1.91	2.61
	$R P E_{t}$	0.07	0.06	0.11	0.12	0.04	0.06	0.10	0.08	0.04	0.10	1.07	0.75
	$R P E_{R}$	0.39	0.48	0.32	1.66	0.25	0.27	0.63	0.35	0.36	0.28	3.76	0.67
	2D-TP	42.62	33.33	75.17	59.79	62.50	89.39	53.68	65.32	24.05	63.18	11.19	44.70
Ours	ATE	0.57	0.61	3.38	0.81	0.09	0.57	0.81	0.66	0.15	0.38	0.25	2.54
	$R P E_{t}$	0.07	0.11	0.33	0.40	0.03	0.08	0.13	0.09	0.04	0.06	0.09	0.42
	$R P E_{R}$	0.28	0.38	0.46	0.40	0.12	0.27	0.37	0.55	0.46	0.24	0.28	0.48
	2D-TP	46.72	34.00	71.43	56.30	70.31	89.02	49.12	56.07	20.57	63.68	11.19	51.37

Note： The units in the table are as follows. ATE： m， RPE_t： m/frame， RPE_R： (°)/frame， TP/%. The best results are highlighted in bold.

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.