Robot Manipulation Generalization On The

Metrics

Average decrease average across all perturbations

Results

Performance results of various models on this benchmark

		Paper Title
SAM2Act	-4.3	-
RVT	-14.5	RVT: Robotic View Transformer for 3D Object Manipulation
3D Diffuser Actor	-15.6	3D Diffuser Actor: Policy Diffusion with 3D Scene Representations
MVP	-16.3	Masked Visual Pre-training for Motor Control
PerAct	-17.13467282	THE COLOSSEUM: A Benchmark for Evaluating Generalization for Robotic Manipulation
PerAct	-17.3	Perceiver-Actor: A Multi-Task Transformer for Robotic Manipulation
RVT2	-19.5	RVT-2: Learning Precise Manipulation from Few Demonstrations
GENIMA	-41.6	Generative Image as Action Models
R3M	-49.9	R3M: A Universal Visual Representation for Robot Manipulation
ACT	-61.8	Learning Fine-Grained Bimanual Manipulation with Low-Cost Hardware

0 of 10 row(s) selected.