Extended Data Table 3 The detailed performance of RoboVLMs on SimplerEnv

From: What matters in building vision–language–action models for generalist robots

  1. (a) Simulation performance on SimplerEnv-WidowX+Bridge environments. We report both the final success rates (final) along with the sub-task success rates (for example, Grasp Spoon). (b) Simulation performance on SimplerEnv-Google Robot (c) Results on WidowX + Bridge setup for cross-embodiment dataset. (d) Results on WidowX + Bridge setup for cross-embodiment dataset. The detailed performance on SimplerEnv.
Back to article page