The essence of humanoid robot motion control technology hinges on its adaptability to diverse environments and proficiency in executing tasks. The Beijing Humanoid Robot Innovation Center has unveiled its proprietary 'Tiangong-Lab' motion control algorithm framework, which seamlessly integrates reinforcement learning and imitation learning. This groundbreaking framework allows robots to autonomously adapt and maintain stable locomotion across complex terrains. Additionally, the center has established a virtual robot training ground for comprehensive scene interaction, leveraging high-fidelity digital twin technology to facilitate large-scale intelligent data collection and training. This innovative approach empowers humanoid robots to achieve agile and flexible operations across a wide range of scenarios.