RoboMME

RoboMME is a large-scale standardized benchmark that contains 16 manipulation tasks designed to evaluate and advance vision-language-action models in long-horizon, history-dependent scenarios, focusing on various aspects of memory such as temporal, spatial, object, and procedural memory.

Papers using RoboMME (2)

DIRECT: When and Where Should You Allocate Test-Time Compute in Embodied Planners?2026RoboMME: Benchmarking and Understanding Memory for Robotic Generalist Policies2026