1 paper
Arusa Kanwal, Pablo Valle, Shaukat Ali +1
Vision-Language-Action (VLA) models are increasingly used as generalist robot policies, yet their evaluation still relies largely on static benchmarks that randomly sample task sce…