VLA-JEPA: Enhancing Vision-Language-Action Model with Latent World Model

Benchmark Model Rank Results
robot-manipulation-on-simplerenv-google-robotVLA-JEPA#12Success Rate: 65.2
robot-manipulation-on-libero-plus-zero-shot-transferVLA-JEPA#25Success Rate: 79.5
robot-manipulation-on-libero-objectVLA-JEPA#32Success Rate: 99.6
robot-manipulation-on-simplerenv-widow-xVLA-JEPA#38Average: 0.573
robot-manipulation-on-libero-longVLA-JEPA#40Success Rate: 95.8
robot-manipulation-on-libero-goalVLA-JEPA#53Success Rate: 97.2
robot-manipulation-on-liberoVLA-JEPA#77Success Rate: 97.2
robot-manipulation-on-libero-spatialVLA-JEPA#82Success Rate: 96.2