Table 5.

Compared results on grasping unseen objects from EGAD! in simulation

Methods	P-Deptha$\left(cm\right)\downarrow$	P-Volumea$\left(c{m}^3\right)\downarrow$	Success Rate $(\%)\uparrow$
Fine-tune	0.668	5.526	45.2
IID-Offline	0.562	6.951	73.0
EWC	0.446	3.341	32.2
SI	0.535	2.654	44.6
BufferSize$\left|\mathcal{M}\right|=\mathbf{1}\mathit{K}$
ER	1.123	17.455	51.9
ER-RM	0.964	20.867	55.5
NI-WLb	0.458	6.874	81.7
NI-WL-RDb	0.760	13.237	65.3
NI-WL-RMb	0.805	16.537	69.4
NI-WL-RM-KDb	0.763	5.822	69.0
BufferSize$\left|\mathcal{M}\right|=\mathbf{5}\mathit{K}$
ER	0.677	5.654	65.4
ER-RM	0.448	6.593	78.5
NI-WL	0.486	7.390	76.9
NI-WL-RD	0.513	8.590	77.9
NI-WL-RM	0.645	9.317	73.4
NI-WL-RM-KD	0.536	3.686	76.9
BufferSize$\left|\mathcal{M}\right|=\mathbf{10}\mathit{K}$
ER	0.474	3.092	74.0
ER-RM	0.473	6.326	76.6
NI-WL	0.488	6.942	76.0
NI-WL-RD	0.407	5.310	74.1
NI-WL-RM	0.432	5.912	73.8
NI-WL-RM-KD	0.636	5.181	76.5

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.