RoboMME: Benchmarking and Understanding Memory for Robotic Generalist Policies

In three linesRoboMME is a standardized benchmark for evaluating memory in vision-language-action (VLA) models for long-horizon robotic manipulation. 16 tasks test temporal, spatial, object, and procedural memory. 14 memory-augmented VLA variants built on π0.5 show effectiveness is highly task-dependent.

Summary generated by Claude — human-verified