Humanoid-Bench
Emerging7papers using it
2025first seen
Humanoid-Bench is a dataset/benchmark used to evaluate humanoid robot loco-manipulation tasks, focusing on task completion and motion coherence in the context of reinforcement learning and imitation learning.
Papers using Humanoid-Bench (7)
- Metaworld: Skill Transfer And Composition In A Hierarchical World Model For Grounding High-level InstructionsFastdsac: Unlocking The Potential Of Maximum Entropy RL In High-dimensional Humanoid ControlNon-conflicting Energy Minimization In Reinforcement Learning Based Robot ControlFasttd3: Simple, Fast, And Capable Reinforcement Learning For Humanoid ControlDoublyAware: Dual Planning and Policy Awareness for Temporal Difference Learning in Humanoid LocomotionMetaWorld: Skill Transfer and Composition in a Hierarchical World Model for Grounding High-Level InstructionsScaling Tasks, Not Samples: Mastering Humanoid Control through Multi-Task Model-Based Reinforcement Learning