Topic
embodied question answering
Artificial Intelligence #synthetic ood generation#robust refusal
Semantic Flip: Synthetic OOD Generation for Robust Refusal in Embodied Question Answering and Spatial Localization
The Semantic Flip framework trains a lightweight rejection module on top of frozen vision-language models to detect unanswerable queries in embodied question answering and spatial localization. It synthesizes out-of-distribution pairs by transforming query and video memory, achieving high refusal accuracy without external OOD annotations.
Jun 16, 2026 1 source
Artificial Intelligence #uav#drones
ScoutVLA: New Dual-Expert AI Model Boosts UAV Active Perception for Embodied Question Answering
Researchers introduce ScoutVLA, a vision-language-action model for UAV active perception, achieving 10.48x higher strict success rate and 7.72x higher QA correctness over baselines. The model features a decoupled dual-expert architecture inspired by scout bee waggle dance.
Jun 16, 2026 1 source