RACO: Reliability-Aware Coarse-Goal Optimization for Inspection-Oriented UAV Vision-Language Navigation
2026-08-24 • Robotics
RoboticsArtificial IntelligenceComputer Vision and Pattern Recognition
AI summaryⓘ
The authors look at how drones navigate to inspect specific objects, pointing out that current methods can get stuck following a rough guess that might be wrong. They create a new testing setup called LG-UVI that includes tricky distractors and detailed checks to see if the drone really reaches and inspects the right place. To improve performance, they develop RACO, a system that treats the rough goal as a hypothesis and continually checks and corrects it using object clues. Their approach makes the drone more reliable at arriving correctly and avoiding false confirmations compared to previous methods.
UAVVision-language navigationCoarse-to-fine navigationInspection regionsGoal localizationTarget objectsDistractorsNavigation policiesObject anchorsEvaluation benchmarks
Authors
Sen Wang, Yiming Sun, Jiaxuan He, Pengfei Zhu
Abstract
UAV vision-language navigation (UAV-VLN) is commonly evaluated as goal reaching, but inspection-oriented deployment requires the agent to stop within a valid inspection region and avoid falsely confirming visually or semantically similar distractors. This requirement exposes a key weakness in existing coarse-to-fine UAV-VLN policies: the coarse goal predicted before local refinement is often treated as reliable, although it may drift toward plausible but incorrect object regions and limit the ability of the local stage to recover. To systematically evaluate this problem, we introduce LG-UVI, an object-centric inspection evaluation setting derived from CityNav/CityRefer. LG-UVI extends standard UAV-VLN episodes with target objects, hard distractors, type-aware inspection regions, and diagnostics for inspection-region arrival and object-level confirmation. To address this inspection-oriented setting, we further propose RACO, a reliability-aware adaptive coarse-to-fine navigation framework. Instead of treating the predicted coarse goal as a fixed waypoint, RACO views it as a runtime hypothesis and uses object-level candidate anchors to check and correct coarse localization before Stage 1 and at the Stage 1-to-Stage 2 boundary. RACO also applies scale-adaptive terminal refinement to handle terminal near-miss cases using runtime-observable geometric and anchor-based evidence. Under a unified online evaluation protocol, RACO improves SR over the reproduced HETT baseline by 9.53 and 7.98 percentage points on validation-unseen and test-unseen, respectively. It also improves inspection-region arrival and reduces false verification risk, showing that coarse-goal reliability optimization is an effective complement to existing coarse-to-fine UAV-VLN policies.