HoST (Humanoid Standing-up Control)

该数据集是一个强化学习框架，它可以从零开始学习站立控制，实现了在多种不同姿势下从模拟到现实的稳健转移。该数据集还包括了成功率、脚步移动、动作平滑度和能量消耗等评估指标，为人类oids在不同环境下的站立控制提供了各种设计选择的有效性见解。

This dataset is a reinforcement learning framework that enables learning standing control from scratch, achieving robust sim-to-real transfer across diverse postures. It also includes evaluation metrics such as success rate, foot movement, motion smoothness, and energy consumption, providing insights into the effectiveness of various design choices for standing control of humanoids in different environments.