Dual Latent Memory in Vision-Language-Action Models for Robotic Manipulation

Benchmark Model Rank Results
robot-manipulation-on-libero-90LaMem-VLA#2Success Rate: 97.0
robot-manipulation-on-simplerenv-widow-xLaMem-VLA#12Average: 0.739
robot-manipulation-on-libero-spatialLaMem-VLA#33Success Rate: 98.8
robot-manipulation-on-libero-longLaMem-VLA#41Success Rate: 95.8
robot-manipulation-on-libero-goalLaMem-VLA#54Success Rate: 97.2
robot-manipulation-on-libero-objectLaMem-VLA#60Success Rate: 99.0