Researchers have introduced GreenWAM, a novel world foundation model designed for robotics. This single model functions as both an action-conditioned world model and a policy, capable of operating across 20 diverse robot families, including bimanual arms and humanoids. GreenWAM demonstrates strong performance in world prediction tasks on WorldArena 1.0, outperforming existing models, and also achieves competitive results as a policy on leaderboards like RoboTwin 2.0 and benchmarks such as LIBERO-Plus, CALVIN, and LIBERO. The model has also been successfully adapted to a real humanoid robot for real-time control. AI
IMPACT This unified model approach could streamline robotics development by providing a single system for both world prediction and action control across diverse robot types.
RANK_REASON The item describes a new research paper introducing a novel foundation model for robotics. [lever_c_demoted from research: ic=1 ai=1.0]
Read on Hugging Face Daily Papers →
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
