Robot Manipulation (Pick Coke Can) on SimplerEnv-Fractal Google Robot
100Success RateGeoAlign
Evaluation Results
| Method | Links | |
|---|---|---|
| GeoAlignEvaluation protocol=Controlled, Source context=ours2026.06 | 100 | |
| RGB-only baselineEvaluation protocol=Controlled, Source context=ours2026.06 | 97.5 | |
| CogACTEvaluation protocol=Public-context (Transcribed), Source context=CogACT Table 1 [49]2026.06 | 91.3 | |
| SpatialVLAEvaluation protocol=Public-context (Transcribed), Source context=MolmoAct Tbl. 1 [38]2026.06 | 81 | |
| RT-2-XEvaluation protocol=Public-context (Transcribed), Source context=MolmoAct Tbl. 1 [38]2026.06 | 78.7 | |
| MolmoActEvaluation protocol=Public-context (Transcribed), Source context=MolmoAct Tbl. 1 [38], Mode=fine-tuned2026.06 | 77.7 | |
| MolmoActEvaluation protocol=Public-context (Transcribed), Source context=MolmoAct Tbl. 1 [38], Mode=zero-shot2026.06 | 71.3 | |
| GR00T-N1.5Evaluation protocol=Public-context (Transcribed), Source context=MolmoAct Tbl. 1 [38]2026.06 | 69.3 | |
| RT-1-XEvaluation protocol=Public-context (Transcribed), Source context=MolmoAct Tbl. 1 [38]2026.06 | 56.7 | |
| MagmaEvaluation protocol=Public-context (Transcribed), Source context=MolmoAct Tbl. 1 [38]2026.06 | 56 | |
| OpenVLAEvaluation protocol=Public-context (Transcribed), Source context=MolmoAct Tbl. 1 [38]2026.06 | 16.3 |