MM-ACT: Learn from Multimodal Parallel Generation to Act

Benchmark Model Rank Results
robot-manipulation-on-libero-objectMM-ACT (Vanilla)#41Success Rate: 99.4
robot-manipulation-on-libero-spatialMM-ACT (Vanilla)#56Success Rate: 97.8
robot-manipulation-on-libero-longMM-ACT (+Text in Long)#68Success Rate: 93.0
robot-manipulation-on-libero-goalMM-ACT (Vanilla)#82Success Rate: 94.8
robot-manipulation-on-liberoMM-ACT (+Text in Long)#94Success Rate: 96.3