Reproduce “H^2MARL: Hierarchical Heterogeneous Multi-agent Reinforcement Learning for Humanoid Robot Control” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Xuemin Hu, Kaixin Li, Qingqing Xia, Yuxin Wang, Wei Xie, Bo Tang, Long Chen
H²MARL coordinates simulated humanoid walking by assigning each torque-controlled joint degree of freedom an independent policy informed by physical properties, while co-optimizing a pose planner and temporal executor through shared closed-loop trajectories. In MuJoCo Humanoid-v2, it improves foot slip, torso stability, and gait symmetry over MAPPO, but its motion-smoothness score is worse; tests are limited to simulation.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.