MemoryBench

MemoryBench is a benchmark that contains a split of memory-dependent tasks from the ClevrSkills environment, used to evaluate the performance of vision-language-action models in handling non-markovian manipulation tasks.

Papers using MemoryBench (3)

MemoAct: Atkinson-Shiffrin-Inspired Memory-Augmented Visuomotor Policy for Robotic Manipulation2026Notes-to-Self: Scratchpad Augmented VLAs for Memory Dependent Manipulation Tasks2026SAM2Act: Integrating Visual Foundation Model with A Memory Architecture for Robotic Manipulation2025