Multi-task Robotic Manipulation on RLBench benchmark leaderboard · SOTA2 Research

Multi-task Robotic Manipulation on RLBench

83.6Avg Success RateVERM

Evaluation Results

Method	Links
VERM2025.12		83.6	96.8	89.6	95.2	100	98.4	99.2	100	99.2	—	80.8	93.6	100	96	55.2	47.2	100	30.4	40
3DDA2025.12		83.5	96	89.6	84	96.8	99.2	97.6	96	100	—	68.3	82.4	97.6	93.6	85.6	44	98.4	65.6	24
FlowRAM2025.06		82.3	85	90	88	82	86.3	93.3	88	100	80.3	31.3	—	—	—	—	—	—	—	—
RVT-22025.12		82.2	100	74	100	99	99	92	96	99	—	80	88	96	95	66	35	100	40	38
3D Diffuser Actor2025.06		78.4	82.7	89.3	94.7	88	80	92	77.3	98.7	69.3	12	—	—	—	—	—	—	—	—
RVT-22025.06		76.2	79.3	78.7	87.3	86.7	85.3	76.7	86.7	96	67.3	27.7	—	—	—	—	—	—	—	—
Act3D2025.12		68.4	93	93	92	94	94	93	90	92	—	12	47	95	80	51	8	99	27	3
Act3D2025.06		65.3	52	84	80	66.7	64	100	54.7	86.7	64	0	—	—	—	—	—	—	—	—
RVT2025.12		65	52	71.2	72	88	93.6	81.6	88	99.2	—	28.8	48	91.2	91	49.6	36	100	11.2	4
PerAct2025.12		52.3	55.2	88	52	70.4	88	74	51.2	89.6	—	26.4	17.6	86	44.8	28	16.8	92.8	5.6	2.4
PolarNet2025.12		48.7	36	84	52	100	80	56	32	92	—	4	44	84	40	12	12	96	4	0
HiveFormer2025.12		48	52	52	28	100	80	64	68	76	—	8	8	76	80	32	8	84	0	0
ManiGaussian2025.06		44.8	28	76	60	60	56	24	16	92	20	12	—	—	—	—	—	—	—	—
GNFactor2025.06		31.7	25.3	76	28	57.3	50.7	20	0	37.3	18.7	4	—	—	—	—	—	—	—	—
PerAct2025.06		20.4	18.7	54.7	0	40	38.7	18.7	2.7	5.3	18.7	6.7	—	—	—	—	—	—	—	—
C2F-ARM-BC2025.12		16.9	24	20	0	20	68	16	4	24	—	0	8	12	8	0	8	72	4	0