RAP
Best current no-additional-threat rate despite the highest displacement error.
FluidTestFluidTest Safety Arena Gold
Human-reviewed planner leaderboard.
FluidTest is currently under an early warm-up stage, this is a temporary leaderboard as indicative results reflecting some SOTA methods measured by RFS from the official WOD-E2E benchmark. Submit to challenge!
Current methods
val151 results table
Higher is better for no-violation rates. 5sADE is lower-is-better.
| Rank | Method | NATR ↑ | Quality ↑ | Nav. ↑ | Colli. ↑ | Long. ↑ | Prior. ↑ | Lateral ↑ | 5sADE ↓ |
|---|---|---|---|---|---|---|---|---|---|
| 1 | RAPRAP research planner | 0.49 | 0.71 | 0.99 | 0.91 | 0.99 | 0.88 | 0.98 | 2.62 |
| 2 | VMAVMA internal planner | 0.47 | 0.72 | 1.00 | 0.84 | 1.00 | 0.93 | 0.97 | 2.46 |
| 3 | PoutinePoutine VLT planner | 0.35 | 0.67 | 0.99 | 0.83 | 0.99 | 0.90 | 0.97 | 2.44 |
VMA
Highest quality and priority scores, second overall on NATR.
Poutine
Lowest 5sADE, but the current table shows the weakest NATR.