Robot Manipulation on RLBench

Metric: Succ. Rate (18 tasks, 100 demo/task) (higher is better)

LeaderboardDataset

Loading chart...

Results

Sort:

#	Model↕	Succ. Rate (18 tasks, 100 demo/task)▼	Extra Data	Paper	Date↕	Code
1	EquAct	89.4	No	-	-	-
2	SAM2Act	86.8	No	-	-	Code
3	ARP+	84.9	No	Autoregressive Action Sequence Learning for Robo...	2024-10-04	Code
4	3D-LOTUS	83.1	No	Towards Generalizable Vision-Language Robotic Ma...	2024-10-02	Code
5	RVT-2	81.4	No	RVT-2: Learning Precise Manipulation from Few De...	2024-06-12	Code
6	3D Diffuser Actor	81.3	No	3D Diffuser Actor: Policy Diffusion with 3D Scen...	2024-02-16	Code
7	Mini Diffuser	77.6	No	Mini Diffuser: Fast Multi-task Diffusion Policy ...	2025-05-14	Code
8	SAM-E	70.6	No	SAM-E: Leveraging Visual Foundation Model with S...	2024-05-30	-
9	Act3D	65	No	Act3D: 3D Feature Field Transformers for Multi-T...	2023-06-30	Code
10	RVT	62.9	No	RVT: Robotic View Transformer for 3D Object Mani...	2023-06-26	Code
11	PerAct (Evaluated in RVT)	49.4	No	Perceiver-Actor: A Multi-Task Transformer for Ro...	2022-09-12	Code
12	PolarNet	46.4	No	PolarNet: 3D Point Clouds for Language-Guided Ro...	2023-09-27	Code
13	Hiveformer	45.3	No	Instruction-driven history-aware policies for ro...	2022-09-11	Code
14	PerAct	42.7	No	Perceiver-Actor: A Multi-Task Transformer for Ro...	2022-09-12	Code
15	C2FARM-BC (Evaluated in PerAct)	20.1	No	Coarse-to-Fine Q-attention: Efficient Learning f...	2021-06-23	Code
16	Image-BC VIT	1.3	No	Perceiver-Actor: A Multi-Task Transformer for Ro...	2022-09-12	Code
17	Image-BC CNN	1.3	No	Perceiver-Actor: A Multi-Task Transformer for Ro...	2022-09-12	Code

#1EquAct
89.4
Succ. Rate (18 tasks, 100 demo/task)
No paper
#2SAM2Act
86.8
Succ. Rate (18 tasks, 100 demo/task)
No paperCode
#3ARP+SOTA
84.9
Succ. Rate (18 tasks, 100 demo/task)· 2024-10-04
Autoregressive Action Sequence Learning for Robotic Manipulation Code
#43D-LOTUSSOTA
83.1
Succ. Rate (18 tasks, 100 demo/task)· 2024-10-02
Towards Generalizable Vision-Language Robotic Manipulation: A Benchmark and LLM-guided 3D Policy Code
#5RVT-2SOTA
81.4
Succ. Rate (18 tasks, 100 demo/task)· 2024-06-12
RVT-2: Learning Precise Manipulation from Few Demonstrations Code
#63D Diffuser ActorSOTA
81.3
Succ. Rate (18 tasks, 100 demo/task)· 2024-02-16
3D Diffuser Actor: Policy Diffusion with 3D Scene Representations Code
#7Mini Diffuser
77.6
Succ. Rate (18 tasks, 100 demo/task)· 2025-05-14
Mini Diffuser: Fast Multi-task Diffusion Policy Training Using Two-level Mini-batches Code
#8SAM-E
70.6
Succ. Rate (18 tasks, 100 demo/task)· 2024-05-30
SAM-E: Leveraging Visual Foundation Model with Sequence Imitation for Embodied Manipulation
#9Act3DSOTA
65
Succ. Rate (18 tasks, 100 demo/task)· 2023-06-30
Act3D: 3D Feature Field Transformers for Multi-Task Robotic Manipulation Code
#10RVTSOTA
62.9
Succ. Rate (18 tasks, 100 demo/task)· 2023-06-26
RVT: Robotic View Transformer for 3D Object Manipulation Code
#11PerAct (Evaluated in RVT)SOTA
49.4
Succ. Rate (18 tasks, 100 demo/task)· 2022-09-12
Perceiver-Actor: A Multi-Task Transformer for Robotic Manipulation Code
#12PolarNet
46.4
Succ. Rate (18 tasks, 100 demo/task)· 2023-09-27
PolarNet: 3D Point Clouds for Language-Guided Robotic Manipulation Code
#13HiveformerSOTA
45.3
Succ. Rate (18 tasks, 100 demo/task)· 2022-09-11
Instruction-driven history-aware policies for robotic manipulations Code
#14PerAct
42.7
Succ. Rate (18 tasks, 100 demo/task)· 2022-09-12
Perceiver-Actor: A Multi-Task Transformer for Robotic Manipulation Code
#15C2FARM-BC (Evaluated in PerAct)SOTA
20.1
Succ. Rate (18 tasks, 100 demo/task)· 2021-06-23
Coarse-to-Fine Q-attention: Efficient Learning for Visual Robotic Manipulation via Discretisation Code
#16Image-BC VIT
1.3
Succ. Rate (18 tasks, 100 demo/task)· 2022-09-12
Perceiver-Actor: A Multi-Task Transformer for Robotic Manipulation Code
#17Image-BC CNN
1.3
Succ. Rate (18 tasks, 100 demo/task)· 2022-09-12
Perceiver-Actor: A Multi-Task Transformer for Robotic Manipulation Code