Sparse-reward manipulation on Threading simulated environment
92Success RateCSIL++ Ens.
Evaluation results
| Method | Links | |
|---|---|---|
| CSIL++ Ens.Policy architecture=Ensemble, Learning step=450k2026.06 | 92 | |
| XQC+OD Res.Policy architecture=Residual, Learning step=0k2026.06 | 60 | |
| CSIL++ Res.Policy architecture=Residual, Learning step=0k2026.06 | 32 | |
| VLA2026.06 | 14 | |
| XQC+OD Ens.Policy architecture=Ensemble, Learning step=150k2026.06 | 4 | |
| PLD RL (Res.)Policy architecture=Residual, Learning step=50k2026.06 | 0 |