Vision-Language Navigation on RxR-CE (val-unseen)
90.4SRHuman
Evaluation results
| Method | Links | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Human2021.10 | 90.4 | 1.32 | — | 77.7 | 74.3 | — | — | — | — | — | |
| Qwen-RobotNav-8BObservation setting=Panoramic2026.06 | 76.5 | 3.58 | 65.7 | 72.5 | — | — | — | — | — | — | |
| Qwen-RobotNav-4BObservation setting=Panoramic2026.06 | 75.2 | 3.8 | 65 | 71.9 | — | — | — | — | — | — | |
| OmniNavTraining Protocol=Supervised Learning (Training Method)2026.05 | 73.6 | 3.77 | 62 | — | — | — | — | — | — | — | |
| OmniNavObservation setting=Panoramic2026.06 | 73.6 | 3.77 | 62 | — | — | — | — | — | — | — | |
| Qwen-RobotNav-8BObservation setting=Monocular2026.06 | 73.4 | 4.16 | 63.5 | 69.9 | — | — | — | — | — | — | |
| AstraNav-World w/ Diffusion PolicyObservation: Pano.=true, Action Decoder=Diffusion Policy2025.12 | 72.9 | 3.82 | 61.5 | — | — | — | — | — | — | — | |
| AstraNav-WorldObservation setting=Panoramic2026.06 | 72.9 | 3.82 | 61.5 | — | — | — | — | — | — | — | |
| Qwen-RobotNav-4BObservation setting=Monocular2026.06 | 71.3 | 4.15 | 61.5 | 68.6 | — | — | — | — | — | — | |
| Image2NavRGB=✓, FOV=180°2026.07 | 70.7 | 3.74 | 59.1 | 71.8 | — | — | — | — | — | — | |
| AstraNav-World w/ Action FormerObservation: Pano.=true, Action Decoder=Action Former2025.12 | 70.4 | 3.93 | 59.6 | — | — | — | — | — | — | — | |
| SPAN-NavRGB=true2026.03 | 69.7 | 4.2 | 60.1 | 67.9 | — | — | — | — | — | — | |
| SPAN-NavSize=-, Observation: Pano.=true, Observation: Depth=false, Observation: RGB=false, Prediction: Extra=true, Prediction: Action=true, Agent Type=Panoramic RGB Agent2026.05 | 69.7 | 4.2 | 60.1 | — | — | — | — | — | — | — | |
| SPAN-NavTraining Protocol=Supervised Learning (Training Method)2026.05 | 69.7 | 4.2 | 60.1 | 67.9 | — | — | — | — | — | — | |
| AgentVLN-3BS.RGB=true, Odo.=true2026.03 | 69.5 | 3.92 | 61.3 | 74.6 | — | — | — | — | — | — | |
| AgentVLNSize=3B, Observation: Pano.=false, Observation: Depth=true, Observation: RGB=true, Prediction: Extra=false, Prediction: Action=true, Agent Type=Egocentric RGB-D Agent2026.05 | 69.5 | 3.9 | 61.3 | — | — | — | — | — | — | — | |
| CorrectNavObservation: S.RGB=true2025.12 | 69.3 | 4.09 | 63.3 | — | — | — | — | — | — | — | |
| ABot-N0Pano.=true2026.02 | 69.3 | 3.83 | 60 | — | — | — | — | — | — | — | |
| CorrectNavS.RGB=true2025.12 | 69.3 | 4.09 | 63.3 | — | — | — | — | — | — | — | |
| ABot-N0Size=4B, Observation: Pano.=true, Observation: Depth=false, Observation: RGB=false, Prediction: Extra=false, Prediction: Action=true, Agent Type=Panoramic RGB Agent2026.05 | 69.3 | 3.8 | 60 | — | — | — | — | — | — | — | |
| ABot-N0Training Protocol=Supervised Learning (Training Method)2026.05 | 69.3 | 3.83 | 60 | — | — | — | — | — | — | — | |
| ABot-N0Observation setting=Panoramic2026.06 | 69.3 | 3.83 | 60 | — | — | — | — | — | — | — | |
| BEVBertWaypoint Predictor=true, Cur. RGB=true, Panoramic View=true, Depth=true, Odometry=true2025.02 | 68.5 | 4 | — | 69.6 | — | — | — | — | — | — | |
| BEVBertS.RGB=true, Pano.=true, Depth=true, Odo.=false, Waypoint Predictor=true2026.05 | 68.5 | 4 | — | 69.6 | — | — | — | — | — | — | |
| BEVBertObservation=Pano. & Depth & Odom., Waypoint=✓2026.06 | 68.5 | 4 | — | 69.6 | — | — | — | — | — | — | |
| AwareVLNS.RGB=true, Pano.=false, Depth=false, Odo.=false, Waypoint Predictor=false2026.05 | 67.6 | 3.95 | 56.1 | 65.7 | — | — | — | — | — | — | |
| StereoNavSize=3B, Observation: Pano.=false, Observation: Depth=false, Observation: RGB=true, Prediction: Extra=true, Prediction: Action=true, Agent Type=Egocentric RGB Agent, Training Data=Additional external data2026.05 | 67.5 | 4.6 | 52 | — | — | — | — | — | — | — | |
| Efficient-VLNSettings=Supervised Learning, Ground truth spatial annotations=false2026.01 | 67 | 3.88 | 54.3 | 68.4 | — | — | — | — | — | — | |
| Efficient-VLNObservation Encoder=Single RGB, Additional Training Data=true, Hardware=H8002025.12 | 67 | 3.88 | 54.3 | 68.4 | — | — | — | — | — | — | |
| Efficient-VLNLearning Paradigm=Supervised Learning2026.02 | 67 | 3.88 | 54.3 | 68.4 | — | — | — | — | — | — | |
| EfficientVLN-4BS.RGB=true2026.03 | 67 | 3.88 | 54.3 | 68.4 | — | — | — | — | — | — | |
| Efficient-VLNLearning Paradigm=Supervised Learning, Pre-Exploration=N/A2026.03 | 67 | 3.88 | 54.3 | 68.4 | — | — | — | — | — | — | |
| Efficient-VLNSize=4B, Observation: Pano.=false, Observation: Depth=false, Observation: RGB=true, Prediction: Extra=false, Prediction: Action=true, Agent Type=Egocentric RGB Agent2026.05 | 67 | 3.9 | 54.3 | — | — | — | — | — | — | — | |
| EfficientVLNRGB=✓, Depth=-2026.07 | 67 | 3.88 | 54.3 | 68.4 | — | — | — | — | — | — | |
| MARVALTraining protocol=Pre-Explore, Privileged access=Yes2022.10 | 66.5 | 4.19 | — | 72.2 | 59.1 | — | — | — | — | — | |
| SPAN-Nav GeneralizeRGB=true2026.03 | 66.5 | 4.5 | 57 | 67.3 | — | — | — | — | — | — | |
| NavForeseePano=true2025.12 | 66.3 | 4.2 | 53.2 | — | — | — | — | — | — | — | |
| NavForeseeSize=3B, Observation: Pano.=true, Observation: Depth=false, Observation: RGB=false, Prediction: Extra=true, Prediction: Action=true, Agent Type=Panoramic RGB Agent2026.05 | 66.3 | 4.2 | 53.2 | — | — | — | — | — | — | — | |
| Image2NavRGB=✓, FOV=90°2026.07 | 65.1 | 4.03 | 54.7 | 70.5 | — | — | — | — | — | — | |
| MARVALTraining protocol=DAGGER2022.10 | 64.8 | 4.49 | — | 70.8 | 57.5 | — | — | — | — | — | |
| NavFoMSettings=Supervised Learning, Ground truth spatial annotations=false2026.01 | 64.4 | 4.74 | 56.2 | 65.8 | — | — | — | — | — | — | |
| NavFoM (Four views)Pano.=true2026.02 | 64.4 | 4.74 | 56.2 | — | — | — | — | — | — | — | |
| BEVBertLearning Paradigm=Supervised Learning2026.02 | 64.4 | 4.8 | — | 65.4 | — | — | — | — | — | — | |
| NavFoMLearning Paradigm=Supervised Learning2026.02 | 64.4 | 4.74 | 56.2 | 65.8 | — | — | — | — | — | — | |
| NavFomRGB=true2026.03 | 64.4 | 4.74 | 56.2 | 65.8 | — | — | — | — | — | — | |
| NavFoMLearning Paradigm=Supervised Learning, Pre-Exploration=N/A2026.03 | 64.4 | 4.74 | 56.2 | 65.8 | — | — | — | — | — | — | |
| NavFoMSize=7B, Observation: Pano.=true, Observation: Depth=false, Observation: RGB=false, Prediction: Extra=false, Prediction: Action=true, Agent Type=Panoramic RGB Agent2026.05 | 64.4 | 4.7 | 56.2 | — | — | — | — | — | — | — | |
| BEVBertLearning paradigm=Supervised learning2026.06 | 64.4 | 4.8 | — | — | — | — | — | — | — | — | |
| NavFoMObservation setting=Panoramic2026.06 | 64.4 | 4.74 | 56.2 | 65.8 | — | — | — | — | — | — | |
| FutureNav-8BVLM Params=8B, Single RGB=true, External Training Data=10435K2026.06 | 63.9 | 4.26 | 54.8 | 69.1 | — | — | — | — | — | — | |
| ACE-Brain-0.5variant=unified2026.07 | 63.8 | 4.3 | 47.9 | 64.6 | — | — | — | — | — | — | |
| MARVALTraining protocol=Finetuned-BC2022.10 | 63.7 | 4.8 | — | 70.6 | 56.9 | — | — | — | — | — | |
| ACE-Brain-0.5-Specialistvariant=specialist, task-specialized tuning=true2026.07 | 63.6 | 4.5 | 51.1 | 67.1 | — | — | — | — | — | — | |
| Efficient-VLNObservation Encoder=Single RGB, Hardware=H8002025.12 | 63.5 | 4.4 | 52.1 | 66.8 | — | — | — | — | — | — | |
| EnvEditEnsemble=3 agents2022.10 | 62.8 | — | — | 68.5 | 54.6 | — | — | — | — | — | |
| SACAPano.=false, Odo.=false, Depth=false, S.RGB=true, Waypoint Predictor=false, Additional training data=true2026.03 | 62.1 | 4.75 | 51.7 | 66 | — | — | — | — | — | — | |
| SACASize=8B, Observation: Pano.=false, Observation: Depth=false, Observation: RGB=true, Prediction: Extra=false, Prediction: Action=true, Agent Type=Egocentric RGB Agent2026.05 | 62.1 | 4.8 | 51.7 | — | — | — | — | — | — | — | |
| Dual-Anchoring Framework (Ours)S-RGB=true2026.04 | 61.7 | 4.42 | 53.3 | — | — | — | — | — | — | — | |
| DualVLNPano.=false, Odo.=false, Depth=false, S.RGB=true, Waypoint predictor=false2025.12 | 61.4 | 4.58 | 51.8 | 70 | — | — | — | — | — | — | |
| DualVLN-7.1BS.RGB=true2026.03 | 61.4 | 4.58 | 51.8 | 70 | — | — | — | — | — | — | |
| DualVLNS-RGB=true2026.04 | 61.4 | 4.58 | 51.8 | — | — | — | — | — | — | — | |
| DualVLNSize=8B, Observation: Pano.=false, Observation: Depth=false, Observation: RGB=true, Prediction: Extra=false, Prediction: Action=true, Agent Type=Egocentric RGB Agent2026.05 | 61.4 | 4.6 | 51.8 | — | — | — | — | — | — | — | |
| DualVLNObservation setting=Monocular2026.06 | 61.4 | 4.58 | 51.8 | 70 | — | — | — | — | — | — | |
| DualVLNRGB=✓, Depth=-2026.07 | 61.4 | 4.58 | 51.8 | 70 | — | — | — | — | — | — | |
| FutureNav-4BVLM Params=4B, Single RGB=true, External Training Data=10457K2026.06 | 60.9 | 4.6 | 52.8 | 68 | — | — | — | — | — | — | |
| SACAPano.=false, Odo.=false, Depth=false, S.RGB=true, Waypoint Predictor=false, Additional training data=false2026.03 | 60.3 | 4.9 | 49.8 | 62.1 | — | — | — | — | — | — | |
| ETP-R1 (Ours-GRPO)training=GRPO-based online RFT2025.12 | 59.92 | 5.22 | 48.97 | 65.31 | 50.41 | — | — | — | — | — | |
| ETP-R1Size=-, Observation: Pano.=true, Observation: Depth=true, Observation: RGB=false, Prediction: Extra=false, Prediction: Action=true, Agent Type=Panoramic RGB-D Agent2026.05 | 59.9 | 5.2 | 49 | — | — | — | — | — | — | — | |
| Qwen-VLA-Instruct2026.07 | 59.6 | 5.8 | 47.8 | 57.1 | — | — | — | — | — | — | |
| MARVALTraining protocol=Pretrained2022.10 | 59.4 | 5.56 | — | 67 | 52.7 | — | — | — | — | — | |
| HSAN2026.06 | 59 | 3.76 | 54 | — | — | 66 | — | — | — | — | |
| SpaAct-stage2Pano.=false, Odo.=false, Depth=false, S.RGB=true, External Data=10476K2026.04 | 58.6 | 5.82 | 49.1 | 64.3 | — | — | — | — | — | — | |
| ETP-R1 (Ours-DAgger)training=online SFT2025.12 | 58.26 | 5.42 | 48.19 | 63.78 | 48.53 | — | — | — | — | — | |
| LatentPilotPano=false, Odo=false, D=false, SRGB=true2026.03 | 58.2 | 5.19 | 49.9 | 67.5 | — | — | — | — | — | — | |
| P3Nav2026.03 | 58.01 | 5.42 | 47.92 | 64.29 | 48.04 | — | — | — | — | — | |
| P³NavSize=-, Observation: Pano.=true, Observation: Depth=true, Observation: RGB=false, Prediction: Extra=false, Prediction: Action=true, Agent Type=Panoramic RGB-D Agent2026.05 | 58 | 5.4 | 47.9 | — | — | — | — | — | — | — | |
| StereoNavSize=3B, Observation: Pano.=false, Observation: Depth=false, Observation: RGB=true, Prediction: Extra=true, Prediction: Action=true, Agent Type=Egocentric RGB Agent, Training Data=Standard navigation data only2026.05 | 58 | 5.9 | 43.5 | — | — | — | — | — | — | — | |
| NavFoMVenue=arXiv2509, VLM Parameters=7B, Observation: Pano.=false, Observation: Odo.=false, Observation: Depth=false, Observation: S.RGB=true2026.01 | 57.4 | 5.51 | 49.4 | 60.2 | — | — | — | — | — | — | |
| NAVIDAVLM Parameters=3B, Observation: Pano.=false, Observation: Odo.=false, Observation: Depth=false, Observation: S.RGB=true2026.01 | 57.4 | 5.23 | 49.6 | 67 | — | — | — | — | — | — | |
| NavFoMObservation Encoder=Single RGB, Additional Training Data=true, Hardware=H1002025.12 | 57.4 | 5.51 | 49.4 | 60.2 | — | — | — | — | — | — | |
| NavFoM-7BS.RGB=true2026.03 | 57.4 | 5.51 | 49.4 | 60.2 | — | — | — | — | — | — | |
| NavFoMS-RGB=true2026.04 | 57.4 | 5.51 | 49.4 | — | — | — | — | — | — | — | |
| NavFoMSize=7B, Observation: Pano.=false, Observation: Depth=false, Observation: RGB=true, Prediction: Extra=false, Prediction: Action=true, Agent Type=Egocentric RGB Agent2026.05 | 57.4 | 5.5 | 49.4 | — | — | — | — | — | — | — | |
| NaVIDASize=3B, Observation: Pano.=false, Observation: Depth=false, Observation: RGB=true, Prediction: Extra=false, Prediction: Action=true, Agent Type=Egocentric RGB Agent2026.05 | 57.4 | 5.2 | 49.6 | — | — | — | — | — | — | — | |
| NavFoMTraining Protocol=Supervised Learning (Training Method)2026.05 | 57.4 | 5.51 | 49.4 | 60.2 | — | — | — | — | — | — | |
| NavFoMVLM Params=7B, Single RGB=true, External Training Data=∼10480K2026.06 | 57.4 | 5.51 | 49.4 | 60.2 | — | — | — | — | — | — | |
| NavFoM2026.07 | 57.4 | 5.5 | 49.4 | 60.2 | — | — | — | — | — | — | |
| NavFoMRGB=✓, Depth=-2026.07 | 57.4 | 5.51 | 49.4 | 60.2 | — | — | — | — | — | — | |
| G3D-LF2026.03 | 57.1 | 5.47 | 47.25 | 63.88 | 47.61 | — | — | — | — | — | |
| HSANAverage of three runs=true2026.06 | 57 | — | 52 | — | — | 62 | 4.7 | — | — | — | |
| HAMT2022.10 | 56.5 | — | — | 63.1 | 48.3 | — | — | — | — | — | |
| HNRSettings=Supervised Learning, Ground truth spatial annotations=false2026.01 | 56.4 | 5.51 | 46.7 | 63.6 | — | — | — | — | — | — | |
| HNRLearning Paradigm=Supervised Learning2026.02 | 56.4 | 5.51 | 46.7 | 63.6 | — | — | — | — | — | — | |
| HNR2024.04 | 56.39 | 5.51 | 46.73 | 63.56 | 47.24 | — | — | — | — | — | |
| HNR2024.10 | 56.39 | 5.51 | 46.73 | 63.56 | 47.24 | — | — | — | — | — | |
| HNR2024.10 | 56.39 | 5.51 | 46.73 | 63.56 | 47.24 | — | — | — | — | — | |
| HNR2025.12 | 56.39 | 5.51 | 46.73 | 63.56 | 47.24 | — | — | — | — | — | |
| HNR-VLN2026.03 | 56.39 | 5.51 | 46.73 | 63.56 | 47.24 | — | — | — | — | — | |
| HNRTraining=Scheduled Sampling2026.07 | 56.39 | 5.51 | 46.73 | 63.56 | 47.24 | — | — | — | — | — | |
| HNRWaypoint Predictor=true, Cur. RGB=true, Panoramic View=true, Depth=true, Odometry=true2025.02 | 56.3 | 5.5 | 46.7 | 63.5 | — | — | — | — | — | — |