Meta-World

Meta-World is a benchmark that contains a diverse set of robotic manipulation tasks used to evaluate the performance of Vision-Language-Action models.

Papers using Meta-World (10)

Multi-View Video Diffusion Policy: A 3D Spatio-Temporal-Aware Video Action Model2026MVR: Multi-view Video Reward Shaping for Reinforcement Learning2026Improving Robustness of Vision-Language-Action Models by Restoring Corrupted Visual Inputs2026MARVL: Multi-Stage Guidance for Robotic Manipulation via Vision-Language Models2026Evo-1: Lightweight Vision-Language-Action Model with Preserved Semantic Alignment2025Self-improving Loops For Visual Robotic Planning2025Pixel Motion Diffusion Is What We Need For Robot Control2025Revisiting The Learning Objectives Of Vision-language Reward Models2025Masked Generative Policy For Robotic Control2025VITA: Zero-Shot Value Functions via Test-Time Adaptation of Vision-Language Models2025