VANGUARD: Vehicle-Anchored Ground Sample Distance Estimation for UAVs in GPS-Denied Environments

Researchers developed VANGUARD, a deterministic tool that helps autonomous drones estimate ground sample distance in GPS-denied environments by using vehicles as reference points. The system addresses critical safety issues with AI vision models that showed over 50% errors in spatial scale estimation, achieving 6.87% median error on benchmark tests.

• →State-of-the-art vision language models suffer from spatial scale hallucinations with median area estimation errors exceeding 50%
• →VANGUARD uses small vehicles as environmental anchors to recover Ground Sample Distance for UAVs without GPS access
• →The system achieved 6.87% median GSD error on the DOTA v1.5 benchmark across 306 images
• →Integration with SAM-based segmentation yielded 19.7% median error with 4x fewer catastrophic failures than VLM baselines
• →The research highlights the need for deterministic geometric tools to ensure safe autonomous spatial reasoning in AI systems