Instruction-following navigation on Real-world Scenes
0.7Success 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 | 0.7 | 0.65 | 0.55 | 0.75 | 0.75 | 0.35 | 0.625 | |
| SparseVideoNavmode=zero-shot2026.02 | 0.55 | 0.55 | 0.5 | 0.65 | 0.5 | 0.25 | 0.5 | |
| Ablation (b) w distilled 4 steps cont. 10mode=zero-shot2026.02 | 0.5 | 0.4 | 0.35 | 0.45 | 0.3 | 0.2 | 0.367 | |
| StreamVLNmode=zero-shot2026.02 | 0.45 | 0.4 | 0.3 | 0.5 | 0.2 | 0.25 | 0.35 | |
| Ablation (d) SparseVideoNav w/o Formermode=zero-shot2026.02 | 0.4 | 0.6 | 0.4 | 0.6 | 0.45 | 0.25 | 0.45 | |
| InternVLA-N1mode=zero-shot2026.02 | 0.25 | 0.15 | 0.35 | 0.3 | 0 | 0 | 0.175 | |
| Ablation (a) w distilled 4 steps cont. 2mode=zero-shot2026.02 | 0.25 | 0.2 | 0.1 | 0.35 | 0.05 | 0 | 0.158 | |
| Uni-NaVidmode=zero-shot2026.02 | 0.2 | 0.1 | 0.1 | 0.2 | 0 | 0 | 0.1 |