Robot Navigation on DynaNav
8.61Navigation ErrorNavDP
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| NavDPNavigation guidance type=Goal point-based2026.02 | 8.61 | 54.12 | 52.62 | 30.59 | |
| BC PolicyNavigation guidance type=Goal point-based, Training strategy=Behavior Cloning2026.02 | 9.96 | 45.88 | 41.52 | 35.29 | |
| TIC-VLAInference synchronicity=Async, Training strategy=RL fine-tuning, Navigation guidance type=Vision-Language2026.02 | 10.55 | 55.29 | 50.29 | 28.24 | |
| TIC-VLA (no RL)Inference synchronicity=Async, Training strategy=BC, Navigation guidance type=Vision-Language2026.02 | 10.85 | 47.06 | 42.41 | 34.12 | |
| RL PolicyNavigation guidance type=Goal point-based, Training strategy=Reinforcement Learning2026.02 | 12.2 | 30.59 | 28.45 | 36.47 | |
| Uni-NaVidNavigation guidance type=Vision-Language2026.02 | 15.9 | 22.35 | 19.61 | 49.41 | |
| TIC-VLA (Sync.)Inference synchronicity=Sync, Navigation guidance type=Vision-Language2026.02 | 16.31 | 32.94 | 29.64 | 41.18 | |
| DualVLNNavigation guidance type=Vision-Language2026.02 | 16.45 | 30.59 | 27.82 | 47.06 | |
| NaVILANavigation guidance type=Vision-Language2026.02 | 17.2 | 28.24 | 25.51 | 48.24 |