OpenVLA: An Open-Source Vision-Language-Action Model

Benchmark Model Rank Results
robot-manipulation-on-simpler-envOpenVLA#6Visual Matching: 0.277Visual Matching-Pick Coke Can: 0.163…
robot-manipulation-on-calvinOpenVLA#19avg. sequence length (D to D): 3.27
robot-manipulation-on-simplerenv-widow-xOpenVLA#49Average: 0.010Put Spoon on Towel: 0.000Put Carrot on Plate: 0.000…
robot-manipulation-on-liberoOpenVLA#163Success Rate: 76.5