Beyond-the-view navigation on Real-world Scenes
35Success Rate (Room)Ablation (c) w/o distilled 50 steps cont. 20
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| Ablation (c) w/o distilled 50 steps cont. 20mode=zero-shot2026.02 | 35 | 45 | 30 | 60 | 20 | 25 | 35.8 | |
| SparseVideoNavmode=zero-shot2026.02 | 25 | 30 | 20 | 40 | 20 | 15 | 25 | |
| Ablation (d) SparseVideoNav w/o Formermode=zero-shot2026.02 | 15 | 25 | 20 | 45 | 20 | 10 | 22.5 | |
| StreamVLNmode=zero-shot2026.02 | 10 | 15 | 5 | 30 | 0 | 0 | 10 | |
| Ablation (b) w distilled 4 steps cont. 10mode=zero-shot2026.02 | 10 | 5 | 5 | 40 | 5 | 5 | 11.7 | |
| Uni-NaVidmode=zero-shot2026.02 | 5 | 0 | 0 | 10 | 0 | 0 | 2.5 | |
| InternVLA-N1mode=zero-shot2026.02 | 5 | 0 | 5 | 40 | 0 | 0 | 8.3 | |
| Ablation (a) w distilled 4 steps cont. 2mode=zero-shot2026.02 | 0 | 0 | 0 | 15 | 0 | 0 | 2.5 |