SafeLIBERO

📊 SafeLIBERO Benchmark 📖 Overview SafeLIBERO is a benchmark designed to evaluate robotic model performance in complex, safety-critical environments. It extends each LIBERO suite by selecting four representative tasks, with each task further divided into two scenarios varying by safety level based on obstacle interfer

Papers using SafeLIBERO (2)

Your Model Already Knows: Attention-Guided Safety Filter for Vision-Language-Action Models2026VLSA: Vision-Language-Action Models with Plug-and-Play Safety Constraint Layer2025