LLaViDA

New framework evaluates autonomous driving VLA models across domains

Researchers have developed SSP, an evaluation framework for autonomous driving vision-language-action (VLA) models. SSP matches events across synthetic, simulated, and physical domains to accurately assess model perform…