PIFR GOAT Nested - Success Rate: leaderboard
PIFR v1 evaluates embodied navigation, with success defined as reaching within 0.25 m geodesic distance of the target. SR and SPL average atomic navigation segments; InterruptSR and ResumeSR use their corresponding stage segments. SG3D, OVON, and GOAT populations remain separate, as do Single, Repeated, and Nested interruption protocols. PIFR-Rank is trained independently for each dataset for 5 epochs. Published system names are retained without mapping them to backbone checkpoints. Table 3 uses seen validation data; Table 4 reports an equal-cell macro-average over 9 dataset-by-structure cells. Metric: Success Rate (%).
Metric: Success Rate (%). Source: arxiv.org. Saturation forecast: Rough model projection: around 2026. 5 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | PIFR GOAT PIFR-Rank | 49.6 | |
| 2 | PIFR GOAT MTU3D | 44.3 | |
| 3 | PIFR GOAT Uni-NaVid | 22.6 | |
| 4 | PIFR GOAT R2F | 18.8 | |
| 5 | PIFR GOAT E-VA | 16.7 |
Interactive version: theaggregate.ai/benchmark?slug=pifr-goat-nested-success-rate · How It Works · Data refreshed daily, snapshot 2026-10-28.