Safety value learning on G1FLAT real hardware
84.42RtempSupervised
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Supervised2026.06 | 84.42 | 27 | 19.92 | |
| lambda-Reachabilitylambda=0.52026.06 | 84.26 | 24 | 18.77 | |
| lambda-Reachabilitylambda=0.952026.06 | 82.24 | 21 | 20.31 | |
| lambda-Reachabilitylambda=ours2026.06 | 80.91 | 19 | 20.31 | |
| lambda-Reachabilitylambda=0.02026.06 | 74.25 | 26 | 31.8 | |
| DPE2026.06 | 62.07 | 44 | 45.21 |