Singapore-based physical artificial intelligence company Ropedia has introduced HOMIE Gen2, a new-generation multimodal data acquisition system. This fourth-generation hardware device, developed over 12 months, marks a new stage of 'scaling experience' in embodied intelligence. Traditional AI suffers from a structural contradiction between digital data and physical experience. HOMIE Gen2 establishes a comprehensive experience acquisition system, capable of synchronously recording multidimensional information across multiple devices. It features long battery life, lightweight design, and high precision, making it directly usable for model training. Ropedia has pioneered a new 'non-self-contained acquisition' approach, significantly reducing acquisition costs and improving deployment efficiency. The feasibility of this approach has been validated through its Xperience-10M dataset. From its prototype to HOMIE Gen2, the system has undergone four hardware innovations, with its 'end-to-end' design providing structured experience data. Currently, Ropedia is building a three-tier architecture system comprising hardware acquisition, data processing, and model training, forming a technological closed loop. The Xperience-10M dataset, with its massive scale, sets a new industry standard.