Vision-Language Navigation on RxR (val-unseen)
65.2Success Rate (SR)Uncertainty-Aware Gaussian Map
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| Uncertainty-Aware Gaussian Map2026.05 | 65.2 | 4.2 | — | 65.6 | 53.5 | |
| BEVBert2026.05 | 64.1 | 4.6 | — | 63.9 | 52.6 | |
| SEDualVLNS.RGB=true2026.05 | 63.9 | 4.12 | 52.4 | 72.8 | — | |
| EnvEdit2026.05 | 62.8 | — | — | 68.5 | 54.6 | |
| DualVLNS.RGB=true2026.05 | 61.4 | 4.58 | 51.8 | 70 | — | |
| Qwen-VLA-Instruct2026.05 | 59.6 | 5.8 | 47.8 | 57.1 | — | |
| NaVIDA2026.04 | 57.4 | 5.23 | 49.6 | 67 | — | |
| NaVIDA + LiveVLNTraining-free=true2026.04 | 56.7 | 5.22 | 49.2 | 67.3 | — | |
| UnprunedRetain Tokens=729, Pruning Ratio=0%2026.03 | 56.53 | 5.71 | 47.26 | 64.14 | — | |
| HAMT2026.05 | 56.5 | — | — | 63.1 | 48.3 | |
| Qwen-VLA-Base2026.05 | 55.1 | 6.4 | 45.8 | 56.2 | — | |
| ETPNavPano.=true, Odo.=true, Depth=true, waypoint predictor=true2026.05 | 54.7 | 5.64 | 44.8 | 61.9 | — | |
| PROSPECT (Ours)†Panoramic=false, Odometry=false, Depth=false, Single-view RGB=true, Training Data=Non-MP3D and Extra data2026.03 | 54.6 | 5.7 | 46.2 | 62.1 | — | |
| StreamVLNExtra Navigation Data=true2026.04 | 54.4 | 5.65 | 45.4 | 63.7 | — | |
| StreamVLN + LiveVLNExtra Navigation Data=true, Training-free=true2026.04 | 53.7 | 5.48 | 44.5 | 63 | — | |
| A-MMRRetain Tokens=146, Pruning Ratio=80%2026.03 | 53.36 | 6.16 | 43.64 | 61.02 | — | |
| A-MMRRetain Tokens=218, Pruning Ratio=70%2026.03 | 53.19 | 6.26 | 45.13 | 61.76 | — | |
| StreamVLN†Panoramic=false, Odometry=false, Depth=false, Single-view RGB=true, Training Data=Non-MP3D and Extra data2026.03 | 52.9 | 6.22 | 46 | 61.9 | — | |
| StreamVLNS.RGB=true2026.05 | 52.9 | 6.22 | 46 | 61.9 | — | |
| StreamVLN2026.05 | 52.9 | 6.2 | 46 | 61.9 | — | |
| PROSPECT (Ours)*Panoramic=false, Odometry=false, Depth=false, Single-view RGB=true, Training Data=MP3D + VideoQA only2026.03 | 52.7 | 5.93 | 42.8 | 60.6 | — | |
| LASAR (Ours)trained_with_MindCraft=true2026.05 | 52.1 | — | 47.9 | — | — | |
| VisPrunerRetain Tokens=218, Pruning Ratio=70%2026.03 | 51.34 | 6.43 | 43.32 | 60.55 | — | |
| LASAR (IL)trained_with_MindCraft=false2026.05 | 50.5 | — | 44.3 | — | — | |
| Uni-NaVidtrained_with_MindCraft=true2026.05 | 50.1 | — | 42.3 | — | — | |
| LASAR (IL+QA)trained_with_MindCraft=true2026.05 | 49.5 | — | 43.3 | — | — | |
| NaVILA†Panoramic=false, Odometry=false, Depth=false, Single-view RGB=true, Training Data=Non-MP3D and Extra data2026.03 | 49.3 | 6.77 | 44 | 58.8 | — | |
| NaViLAExtra Navigation Data=true2026.04 | 49.3 | 6.77 | 44 | 58.8 | — | |
| NaVILAtrained_with_MindCraft=true2026.05 | 49.3 | — | 44 | — | — | |
| NaVILAS.RGB=true2026.05 | 49.3 | 6.77 | 44 | 58.8 | — | |
| NaVILA2026.05 | 49.3 | 6.8 | 44 | 58.8 | — | |
| VisPrunerRetain Tokens=146, Pruning Ratio=80%2026.03 | 49.16 | 6.73 | 40.15 | 57.64 | — | |
| Uni-NavidPanoramic=false, Odometry=false, Depth=false, Single-view RGB=true2026.03 | 48.7 | 6.24 | 40.9 | — | — | |
| UniNaVidExtra Navigation Data=true2026.04 | 48.7 | 6.24 | 40.9 | — | — | |
| Uni-NaVidtrained_with_MindCraft=false2026.05 | 48.7 | — | 40.9 | — | — | |
| Uni-NaVid2026.05 | 48.7 | 6.2 | 40.9 | — | — | |
| StreamVLN*Panoramic=false, Odometry=false, Depth=false, Single-view RGB=true, Training Data=MP3D + VideoQA only2026.03 | 48.6 | 6.72 | 42.5 | 60.2 | — | |
| StreamVLN2026.04 | 48.6 | 6.72 | 42.5 | 60.2 | — | |
| A-MMRRetain Tokens=72, Pruning Ratio=90%2026.03 | 45.71 | 6.9 | 32.91 | 47.69 | — | |
| NaVid2026.05 | 45.7 | 5.7 | 38.2 | — | — | |
| DivPruneRetain Tokens=218, Pruning Ratio=70%2026.03 | 44.48 | 7.29 | 32.15 | 50.62 | — | |
| EnvDrop+2026.05 | 42.6 | — | — | 55.7 | — | |
| SparseVLMRetain Tokens=218, Pruning Ratio=70%2026.03 | 40.87 | 7.18 | 35.02 | 53.01 | — | |
| VisPrunerRetain Tokens=72, Pruning Ratio=90%2026.03 | 37.34 | 7.6 | 25.34 | 41.05 | — | |
| DivPruneRetain Tokens=146, Pruning Ratio=80%2026.03 | 35.13 | 8.49 | 23.92 | 42.68 | — | |
| SparseVLMRetain Tokens=146, Pruning Ratio=80%2026.03 | 34.59 | 7.64 | 30.1 | 52.89 | — | |
| NavMorphPanoramic=false, Odometry=false, Depth=true, Single-view RGB=true2026.03 | 30.8 | 8.85 | 22.8 | 44.2 | — | |
| VLN⟲BERTPanoramic=true, Odometry=true, Depth=true, Single-view RGB=false2026.03 | 27 | 8.98 | 22.6 | 46.7 | — | |
| BERTPano.=true, Odo.=true, Depth=true, waypoint predictor=true2026.05 | 27 | 8.98 | 22.6 | 46.7 | — | |
| CMAPanoramic=true, Odometry=true, Depth=true, Single-view RGB=false2026.03 | 26.5 | 8.76 | 22.1 | 47 | — | |
| CMAPano.=true, Odo.=true, Depth=true, waypoint predictor=true2026.05 | 26.5 | 8.76 | 22.1 | 47 | — | |
| Navidtrained_with_MindCraft=true2026.05 | 26.3 | — | 22.8 | — | — | |
| NavCoTtrained_with_MindCraft=true2026.05 | 26.2 | — | 18.7 | — | — | |
| ETPNav+FFPanoramic=false, Odometry=true, Depth=true, Single-view RGB=true2026.03 | 25.5 | 8.79 | 18.1 | — | — | |
| NavCoTtrained_with_MindCraft=false2026.05 | 24.5 | — | 22.6 | — | — | |
| DivPruneRetain Tokens=72, Pruning Ratio=90%2026.03 | 23.22 | 9.83 | 14.56 | 33.73 | — | |
| SparseVLMRetain Tokens=72, Pruning Ratio=90%2026.03 | 23.17 | 8.51 | 20.87 | 50.06 | — | |
| LSTM2026.05 | 22.8 | 10.9 | — | 38.9 | 18.2 | |
| Seq2SeqPanoramic=false, Odometry=false, Depth=true, Single-view RGB=true2026.03 | 13.9 | 12.1 | 11.9 | 30.8 | — | |
| Seq2SeqOdo.=true, Depth=true2026.05 | 13.9 | 12.1 | 11.9 | 30.8 | — | |
| LAWPanoramic=false, Odometry=true, Depth=false, Single-view RGB=true2026.03 | 8 | 10.9 | 8 | 38 | — | |
| LAWPano.=true, Odo.=true, Depth=true2026.05 | 8 | 10.9 | 8 | 38 | — |