- Rating
- 3(+1)
- Confidence
- 2
- Informativeness
- 2
- Presentation
- 2
- Soundness
- 2
Read review
Paper summary
This paper addresses UAV vision-language navigation and argues that the evidence an agent needs changes once the target has been found. Before discovery, the agent explores using scene context and the instruction. After discovery, the task becomes target-centered local control, split into three sub-stages: approach, regrounding, and reacquisition. The method identifies the target with an open-vocabulary detector and a vision-language verifier, and it keeps a lightweight target memory to decide the current stage. It then routes only the visual and language inputs suited to that stage into a single shared action policy. Experiments on the TravelUAV benchmark show solid improvements over existing methods, including on unseen objects and unseen maps.
Critical strengths
- The motivation is clear and intuitive: once a static target is grounded, the task reduces to target-centered local control.
- The stage logic is explicit and interpretable (Table 1). Separating presence verification from ranking, and allowing landing only when the target is currently localized, are sound design choices.
- Gains are consistent across all three splits in the Full setting. The controlled comparison in Table 6 directly supports stage-adaptive routing.
- The appendix gives detailed training data, schedules, and thresholds, which helps reproducibility.
Critical weaknesses
- Heavy reliance on the oracle directional hint. At every step, including test time, the hint is computed from the ground-truth target position (Eq. 1), as the UAV-Need-Help protocol allows. Beyond the policy input, the method builds view routing and 10° geometric candidate filtering on top of this hint, so its target identification may depend on it much more than the baselines do. The paper doesn't say whether baselines were retrained to use the hint equally well. Results with a noisy or coarser hint would show how robust the method really is.
- The efficiency claim isn't supported. The paper says uniform multimodal fusion adds unnecessary computation. Yet each step chains Grounding DINO, up to three Qwen3-VL-8B calls, and OpenVLA-7B, and the limitations section admits latency goes up. Per-step latency, memory, and parameter comparisons with the baselines should be reported.
- The experimental analysis is incomplete. The method falls behind AerialVLA on some subsets, such as NE and OSR on Unseen Map - Easy and SPL on Seen Hard, and these aren't discussed. Stage-determination accuracy and a failure breakdown are missing, so it's hard to tell whether errors come from perception or control. No variance across seeds is reported, which matters for the +3.13 SR gain on Unseen Map.
- No real-world experiments. Everything is evaluated in simulation, with no physical UAV deployment or sim-to-real validation. The approach stage drops background and language inputs and assumes a clear line of sight to a static target, so robustness to obstacles, occlusion, wind, and sensor noise is untested. At least a small real-world or perturbation study is needed.
Questions for authors
N/A
6 comments
Response part 2
| Decision stage | Step proportion | Semantic verification | Candidate ranking | Stage comparison | Stage-decision subtotal | Action policy | Complete decision |
|---|---|---|---|---|---|---|---|
| Exploration | 32.25% | 0.755 s | 0.095 s | 0.361 s | 1.211 s | 0.677 s | 1.888 s |
| Approach | 20.33% | 3.173 s | 1.305 s | 2.605 s | 7.083 s | 0.670 s | 7.753 s |
| Regrounding | 19.95% | 4.037 s | 1.251 s | 3.724 s | 9.012 s | 0.663 s | 9.675 s |
| Reacquisition | 27.47% | 0.743 s | 0.105 s | 0.385 s | 1.233 s | 0.783 s | 2.016 s |
Our method requires more computation per step, more parameters, and more GPU memory: its stage-weighted complete decision latency is 4.669 s per step, or 2.84 times that of AerialVLA and 2.56 times that of TravelUAV. Action-policy inference remains at 0.663–0.783 s per step across the four stages; the added time mainly comes from semantic verification and stage decisions, especially the more detailed checks in Approach and Regrounding. However, our method averages only 62 navigation steps, versus 169 for AerialVLA and 171 for TravelUAV. Mean cumulative decision time per episode is 289.478 s, only 4.25% above AerialVLA and 7.29% below TravelUAV. Overall navigation efficiency therefore does not decline substantially. Given the improved success rate of the full method, we consider this computational cost acceptable.
W3: The experimental analysis is incomplete. The method trails AerialVLA on some metrics and subsets, including NE and OSR on Unseen Map and SPL on Seen Hard, without discussion. Stage accuracy and failure decomposition are missing, making it hard to distinguish perception from control errors. There is no seed variability analysis, which matters for the +3.13-point SR gain on Unseen Map.
Response: Thank you for your comment.
The original manuscript reported performance on each subset, but did not analyze the local disadvantages or display the multi-seed results obtained earlier. Below, we first list the local differences across the relevant subsets, then report the aggregate results for seeds 41–50 to examine whether the full method's gain relative to AerialVLA is stable.
| Subset and metric | AerialVLA | Our method | Difference (Ours − baseline) |
|---|---|---|---|
| Unseen Map Easy · NE ↓ | 44.99 m | 53.19 m | +8.20 m |
| Unseen Map Easy · OSR ↑ | 58.47% | 57.14% | −1.33 points |
| Unseen Map Easy · SR ↑ | 41.89% | 45.81% | +3.92 points |
| Seen Hard · SPL ↑ | 40.26% | 38.48% | −1.78 points |
| Seen Hard · SR ↑ | 46.30% | 51.39% | +5.09 points |
On Unseen Map Easy, our SR is higher than AerialVLA's, but OSR is slightly lower and mean NE is higher. Our retrospective analysis indicates that unseen-map obstacles can make a coarse direction hint insufficient to identify a passable route; when a similar distractor is mistaken for the target, the policy can approach the wrong location. Both can prevent some trajectories from reaching the true target neighborhood, lowering OSR and raising NE. On Seen Hard, we observed that occlusion-triggered reacquisition can make successful trajectories backtrack, while the limited scene context during Approach can require detours or retreat around obstacles. Both increase flown path length and lower SPL.
To distinguish perception from control further, we checked the stage at each decision step and corrected erroneous stage decisions one at a time on failed trajectories. Each corrected rollout resumes from the same simulator state, replacing only the predicted stage with its ground-truth counterpart while holding target observations, policy weights, and the official success condition fixed. This tests whether stage correction improves navigation.
| Stage | Stage classification accuracy | Description |
|---|---|---|
| Exploration | 83.7% | Global search before target discovery |
| Approach | 81.3% | Local approach after target localization |
| Regrounding | 76.8% | Reverification of current and remembered targets |
| Reacquisition | 70.5% | Recovery after temporary target loss |
| Statistic | Definition | Result |
|---|---|---|
| Fraction of correction points yielding success after ground-truth stage substitution | Correct one point at a time and rerun closed-loop; successful corrections ÷ all tested erroneous transitions | 47.89% |
| NE change at correction points that remained unsuccessful | Paired comparison of final-position error in the original and corrected trajectories | 84.6 m → 72.9 m |
When an erroneous stage decision was replaced with its ground-truth stage from the same simulator state, the originally failed navigation succeeded for 47.89% of the tested correction points. For reruns that still failed, mean NE decreased from 84.6 m to 72.9 m. This shows that correct stage selection can improve subsequent action decisions and closed-loop navigation outcomes. Because changing the stage also changes the evidence received by the action policy, this improvement reflects the joint effect of stage routing and action decisions; their independent contributions cannot be separated. We further report multi-seed results to help assess the robustness of the full method's gain relative to AerialVLA.
Response part 3
To show seed variability, we summarize our results for seeds 41–50 on the same splits and under the same evaluation protocol as mean ± standard deviation, alongside the results reported for AerialVLA. The following tables retain NE, SR, OSR, and SPL from Tables 2–4.
Table 2. Seen split
| Setting | Method | NE (m) ↓ | SR (%) ↑ | OSR (%) ↑ | SPL (%) ↑ |
|---|---|---|---|---|---|
| Full | TravelUAV | 98.66 | 17.45 | 48.87 | 15.76 |
| NavFoM | 93.05 | 29.17 | 49.24 | 25.03 | |
| LongFly | 60.02 | 36.39 | 65.87 | 31.07 | |
| AerialVLA | 65.88 | 47.96 | 57.69 | 38.54 | |
| Ours | 37.96±2.23 | 58.41±1.37 | 79.06±1.55 | 40.62±0.85 | |
| Easy | TravelUAV | 66.40 | 20.26 | 51.23 | 18.10 |
| NavFoM | 58.98 | 32.91 | 53.16 | 27.87 | |
| LongFly | 38.10 | 38.52 | 71.90 | 31.24 | |
| AerialVLA | 43.76 | 49.30 | 61.30 | 37.14 | |
| Ours | 19.87±2.62 | 64.10±2.00 | 85.73±2.14 | 42.36±1.43 | |
| Hard | TravelUAV | 138.04 | 14.02 | 45.98 | 12.90 |
| NavFoM | 143.83 | 23.58 | 43.40 | 20.80 | |
| LongFly | 85.20 | 33.94 | 58.94 | 30.88 | |
| AerialVLA | 93.16 | 46.30 | 53.23 | 40.26 | |
| Ours | 60.26±3.83 | 51.39±1.86 | 70.83±2.39 | 38.48±0.65 |
Table 3. Unseen Object split
Full: n = 629; Easy: n = 209; Hard: n = 420.
| Setting | Method | NE (m) ↓ | SR (%) ↑ | OSR (%) ↑ | SPL (%) ↑ |
|---|---|---|---|---|---|
| Full | TravelUAV | 118.11 | 22.42 | 46.90 | 20.51 |
| NavFoM | 108.04 | 29.83 | 47.99 | 27.20 | |
| LongFly | 66.74 | 43.87 | 64.56 | 38.39 | |
| AerialVLA | 61.45 | 56.60 | 64.86 | 46.61 | |
| Ours | 39.09±2.45 | 66.29±1.46 | 81.83±1.64 | 53.56±1.46 | |
| Easy | TravelUAV | 86.12 | 24.40 | 49.28 | 22.03 |
| NavFoM | 70.51 | 32.54 | 50.72 | 29.54 | |
| LongFly | 54.84 | 38.01 | 56.84 | 31.36 | |
| AerialVLA | 45.72 | 56.94 | 64.11 | 43.76 | |
| Ours | 46.04±0.21 | 62.00±1.83 | 72.90±2.03 | 45.51±0.75 | |
| Hard | TravelUAV | 134.03 | 21.43 | 45.71 | 19.75 |
| NavFoM | 133.01 | 28.03 | 46.18 | 25.64 | |
| LongFly | 57.07 | 50.25 | 74.16 | 45.27 | |
| AerialVLA | 69.27 | 56.43 | 65.24 | 48.03 | |
| Ours | 35.63±3.61 | 68.43±2.06 | 86.27±2.25 | 57.57±2.12 |
Table 4. Unseen Map split
Full: n = 958; Easy: n = 561; Hard: n = 397.
| Setting | Method | NE (m) ↓ | SR (%) ↑ | OSR (%) ↑ | SPL (%) ↑ |
|---|---|---|---|---|---|
| Full | TravelUAV | 138.80 | 4.18 | 20.77 | 3.84 |
| NavFoM | 125.10 | 6.30 | 18.95 | 5.68 | |
| LongFly | 108.32 | 11.27 | 30.27 | 9.32 | |
| AerialVLA | 67.42 | 37.58 | 52.92 | 28.22 | |
| Ours | 54.13±2.45 | 40.71±0.88 | 53.32±0.42 | 31.68±0.94 | |
| Easy | TravelUAV | 102.94 | 4.63 | 22.82 | 4.24 |
| NavFoM | 102.41 | 6.77 | 20.07 | 6.04 | |
| LongFly | 78.56 | 12.96 | 34.31 | 10.32 | |
| AerialVLA | 44.99 | 41.89 | 58.47 | 29.72 | |
| Ours | 53.19±3.20 | 45.81±1.46 | 57.14±0.45 | 33.92±1.53 | |
| Hard | TravelUAV | 189.46 | 3.53 | 17.88 | 3.28 |
| NavFoM | 170.58 | 5.36 | 15.71 | 4.97 | |
| LongFly | 148.10 | 9.02 | 24.88 | 7.98 | |
| AerialVLA | 99.11 | 31.49 | 45.09 | 26.11 | |
| Ours | 55.47±3.82 | 33.50±0.75 | 47.92±0.84 | 28.52±0.80 |
Across the three tables, the full method's mean SR on Seen, Unseen Object, and Unseen Map Full exceeds the published AerialVLA results by 10.45, 9.69, and 3.13 percentage points, respectively. The cross-seed standard deviations shown are 1.37, 1.46, and 0.88 percentage points. Positive SR gains also appear on the Easy and Hard subsets of all three splits. These multi-seed summaries provide supplementary evidence for the robustness of the full method's gain relative to AerialVLA.
W4: There are no real-world experiments. All evaluation is in simulation, with no physical-drone deployment or sim-to-real validation. During Approach, the policy drops background and language inputs and assumes a clear view of a static target. Robustness to obstacles, occlusion, wind, and sensor noise remains untested. At least a small real-world or perturbation study is needed.
Response: Thank you for pointing out the importance of real-world validation. Published UAV-Need-Help comparisons mainly rely on reproducible simulation tasks, and there is not yet a unified real-world evaluation protocol. UAV-Need-Help requires a UAV to search for, recognize, and land near a target that is initially invisible; the task itself is very difficult. Consequently, related methods have conducted their evaluations in simulation. We follow the official simulation splits and success criteria to align with these methods. Following your suggestion, we also added a small physical-drone experiment to test whether stage routing can support closed-loop navigation under real obstacles, occlusion, and wind disturbance.
We used a controlled 28 m × 28 m indoor arena with two kinds of static-target tasks, both with the target initially out of view. In the first, few obstacles surround the target and the drone can approach directly after discovery. The second includes box obstacles and visually similar distractors, so detours can temporarily hide the target. For each fixed layout, we tested ten distinct targets once each, giving 20 trials with a 70-decision-step limit. The platform uses a Jetson Orin NX 16GB, a MicoAir743v2 flight controller, and five D435 RGB-D cameras facing forward, backward, left, right, and down. RGB supports recognition and high-level action prediction; depth and fused pose support local mapping and collision avoidance.
Response part 1
Thank you for your comment and review. We find that navigation decisions require different evidence before and after target discovery: before discovery, they depend on the global scene and the full instruction; afterward, they shift to target-driven local control. We therefore propose multi-stage adaptive evidence routing and lightweight target memory. By selecting visual and language inputs by stage, a shared action policy can handle long-range search, target approach, regrounding, and recovery after target loss. This addresses the difficulty of adapting fixed inputs to every stage and avoids interference from substantial redundant information. Across all 3,005 official TravelUAV test trajectories, our method's SR exceeds that of the strongest baseline reported in the paper on Seen, Unseen Object, and Unseen Map. We respond in detail below to your concerns under Weaknesses and Questions.
W1: The method relies heavily on an oracle direction hint computed from the ground-truth target position at every step, including test time (Eq. 1), as permitted by the UAV-Need-Help protocol. The hint drives not only the policy input but also view routing and the 10° geometric candidate filter, so target recognition may depend on it more than in the baselines. The paper does not say whether baselines were retrained to use the hint equally effectively. Results with noisy or coarse hints would show the method's actual robustness.
Response: Thank you for your comment. Because UAV-Need-Help is a difficult long-range navigation task, all compared methods use the target-direction information provided by the task to assist search. Our method follows the same task protocol and uses relative bearing information without additionally providing target coordinates or a ground-truth path. It receives a quantized, coarse relative bearing rather than target coordinates, distance, or a path. Because the hint also participates in target recognition, we agree that its sensitivity to errors should be tested.
We tested navigation with zero-mean Gaussian noise added to the target's relative bearing at each step, then mapped the perturbed angle to a direction category. Mild, moderate, and severe noise use standard deviations of 10°, 30°, and 60°, respectively. “No Hint” retains the original full task instruction but removes the step-wise dynamic direction hint. The results are:
As direction noise increases, our stage-error rate rises from 21.33% without noise to 27.54%, 40.38%, and 57.64%; removing the dynamic hint raises it to 62.57%. Thus, inaccurate direction information increasingly disrupts stage decisions. At the navigation level, SR and SPL decline for all three methods. Our method retains the highest SR, OSR, and SPL and the lowest NE in all five conditions. Hint quality affects both stage classification and navigation, but our method preserves a relative advantage under the tested perturbations.
W2: The efficiency claim is not fully supported. The paper argues that uniform multimodal fusion adds unnecessary computation, yet each step chains Grounding DINO, up to three Qwen3-VL-8B calls, and OpenVLA-7B. The limitations section acknowledges increased latency. Please compare per-step latency, memory, and parameter counts with baselines.
Response: Thank you for your comment.
To assess efficiency, we measured latency, GPU memory, and parameter counts on the same hardware and compared our method with the released baseline implementations. NavFoM and LongFly were excluded because their code was unavailable. The results are:
We also measured module latencies and step proportions for each stage: