Table 4.

Compared results on grasping seen objects from YCB in simulation

Methods	P-Deptha$\left(cm\right)\downarrow$	P-Volumea$\left(c{m}^3\right)\downarrow$	Success Rate $(\%)\uparrow$
Fine-tune	0.824	9.142	32.8
IID-Offline	0.642	7.273	62.0
EWC	0.714	8.530	23.3
SI	0.733	8.663	33.6
BufferSize$\left|\mathcal{M}\right|=\mathbf{1}\mathit{K}$
ER	1.262	23.204	41.4
ER-RM	0.895	11.928	49.3
NI-WLb	0.516	6.115	54.7
NI-WL-RDb	0.755	10.073	45.3
NI-WL-RMb	0.731	9.963	50.6
NI-WL-RM-KDb	0.818	6.413	47.1
BufferSize$\left|\mathcal{M}\right|=\mathbf{5}\mathit{K}$
ER	0.668	8.021	53.0
ER-RM	0.477	5.211	55.2
NI-WL	0.496	5.868	55.5
NI-WL-RD	0.533	6.587	54.7
NI-WL-RM	0.531	5.908	54.5
NI-WL-RM-KD	0.613	4.400	55.4
BufferSize$\left|\mathcal{M}\right|=\mathbf{10}\mathit{K}$
ER	0.608	6.340	55.4
ER-RM	0.455	4.703	55.9
NI-WL	0.512	5.720	52.7
NI-WL-RD	0.438	4.458	54.6
NI-WL-RM	0.482	5.201	53.1
NI-WL-RM-KD	0.661	4.355	51.3

Bold underline, italic underline, and underline font highlights the first place, second place, and third place with same BufferSize, respectively.

^aP-Depth is short for Penetration Depth, P-Volume is short for Penetration Volume.

^bNI-WL, NI-WL-RD, NI-WL-RM, and NI-WL-RM-KD are four variants of our proposed method.