LaMP: Learning Vision-Language-Action Policies with 3D Scene Flow as Latent Motion Prior

Open paper

Benchmark	Model	Rank	Results
robot-manipulation-on-simplerenv-widow-x	LaMP	#8	Average: 0.792
robot-manipulation-on-libero-plus-zero-shot-transfer	LaMP	#27	Success Rate: 79.3
robot-manipulation-on-libero	LaMP	#45	Success Rate: 98.3