Benchmark RadarAI BENCHMARK PROFILE
AeroGround
AeroGround evaluates vision-language models on aerial-ground collaborative reasoning using a simulated dataset of ~29,000 multimodal observation groups and 2,250 QA instances covering cross-view correspondence, spatial understanding, and reasoning.
- Released
- 2026-08-12
- Readiness
- Paper only
- Primary field
- Robotics & Autonomous Systems
Why it matters
Existing UAV benchmarks focus on aerial-only views; AeroGround fills the gap for aerial-ground collaboration, offering a standardized evaluation for models in real-world applications like rescue and inspection, with clear human performance comparison.
Motivation
Vision-language models (VLMs) have been widely employed in understanding and reasoning tasks for unmanned aerial vehicles (UAVs).
Primary resources
Benchmark Radar records only publicly supported details and links back to primary sources for verification.