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

Open Access

Table 2

Camera and object pose estimation results compared with other visual SLAMMOT methods on the OMD swinging_4_unconstrained sequence

Camera and object	ClusterVO^[14]		VDO-SLAM^[15]		SDPL-SLAM^[21]		DynoSAM^[22]		Ours
Camera and object	$R P E_{t}$	$R P E_{R}$	$R P E_{t}$	$R P E_{R}$	$R P E_{t}$	$R P E_{R}$	$R P E_{t}$	$R P E_{R}$	$R P E_{t}$	$R P E_{R}$
Camera	0.007	0.767	0.017	0.466	0.014	0.453	0.006	0.690	0.014	0.331
Top-left Box	0.067	3.254	0.036	1.494	0.031	1.452	‒	‒	0.015	0.568
Top-right Box	0.026	3.531	0.033	1.369	0.032	1.367	‒	‒	0.010	0.525
Bottom-left Box	0.076	4.915	0.027	1.601	0.027	1.605	‒	‒	0.014	0.522
Bottom-right Box	0.014	4.068	0.030	1.166	0.029	1.164	‒	‒	0.009	0.477
Mean Box	0.046	3.942	0.032	1.407	0.030	1.397	0.040	3.200	0.012	0.523

Note： The units in the table are as follows. RPE_t： m/frame， RPE_R： (°)/frame. 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.