Vision-Language Navigation on OpenUAV test unseen 1.0 (full)
66.74Navigation ErrorLongFly(Ours)
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| LongFly(Ours)spatiotemporal context integration=true, backbone=Qwen2025.12 | 66.74 | 4,387 | 6,456 | 3,839 | |
| SpatialFly2026.03 | 87.82 | 25.46 | 44.41 | 21.76 | |
| LongFly2025.12 | 91.84 | 2,419 | 4,386 | 2,084 | |
| LongFly2026.03 | 91.84 | 24.19 | 43.86 | 20.84 | |
| BS(Ours)spatiotemporal context integration=false, backbone=Qwen2025.12 | 96.13 | 2,384 | 3,783 | 2,096 | |
| SpatialFlyvariant=Ours Full2026.03 | 104.05 | 13.57 | 30.38 | 11.13 | |
| BSDescription=Qwen-based base system without SHIC and STE2025.12 | 106.08 | 1,399 | 2,766 | 1,216 | |
| BS2026.03 | 106.08 | 13.99 | 27.66 | 12.16 | |
| NavFoM2025.12 | 108.04 | 2,983 | 4,799 | 2,720 | |
| LongFly(Ours)backbone=Qwen, spatiotemporal context integration=true2025.12 | 108.32 | 11.27 | 30.27 | 9.32 | |
| LongFly2026.03 | 108.32 | 11.27 | 30.27 | 9.32 | |
| BS(Ours)backbone=Qwen, spatiotemporal context integration=false2025.12 | 112.61 | 7.52 | 20.98 | 6.38 | |
| BSvariant=Ours Baseline2026.03 | 112.61 | 7.52 | 20.98 | 6.38 | |
| TravelUAV2025.12 | 118.11 | 2,242 | 4,690 | 2,051 | |
| NavFoM2025.12 | 118.34 | 1,563 | 3,046 | 1,421 | |
| NavFoM2026.03 | 118.34 | 15.63 | 30.46 | 14.21 | |
| NavFoMtype=navigation foundation model, input=multi-view video and natural language2025.12 | 125.1 | 6.3 | 18.95 | 5.68 | |
| NavFoM2026.03 | 125.1 | 6.3 | 18.95 | 5.68 | |
| TravelUAV2025.12 | 130.6 | 1,141 | 3,113 | 1,045 | |
| TravelUAV2026.03 | 130.6 | 11.41 | 31.13 | 10.45 | |
| TravelUAVplatform=OpenUAV, representation=unified multimodal2025.12 | 138.8 | 4.18 | 20.77 | 3.84 | |
| TravelUAV2026.03 | 138.8 | 4.18 | 20.77 | 3.84 | |
| CMAarchitecture=Cross-Modal Attention, decoder=trajectory decoder2025.12 | 141.68 | 2.3 | 10.02 | 2.16 | |
| CMA2026.03 | 141.68 | 2.3 | 10.02 | 2.16 | |
| CMA2025.12 | 147.27 | 498 | 1,241 | 474 | |
| CMA2026.03 | 147.27 | 4.98 | 12.41 | 4.74 | |
| CMA2025.12 | 155.79 | 906 | 1,606 | 868 | |
| Fixed Actionbehavior=deterministic macro-actions2025.12 | 180.47 | 0.52 | 2.61 | 0.39 | |
| Fixed Action2026.03 | 180.47 | 0.52 | 2.61 | 0.39 | |
| Fixed Action2025.12 | 193.3 | 176 | 536 | 109 | |
| Fixed Action2026.03 | 193.3 | 1.76 | 5.36 | 1.09 | |
| Random Actionsampling=uniformly at random2025.12 | 202.98 | 0 | 0 | 0 | |
| Random Action2026.03 | 202.98 | 0 | 0 | 0 | |
| Fixed Action2025.12 | 212.84 | 366 | 954 | 216 | |
| Random Action2025.12 | 225.64 | 6 | 6 | 6 | |
| Random Action2026.03 | 225.64 | 0.06 | 0.06 | 0.06 | |
| Random Action2025.12 | 260.14 | 16 | 16 | 16 |