📄 中文摘要
大型语言模型经历了显著的转折点，海量文本数据突然赋予了它们意想不到的能力。现在，机器人技术也迎来了类似的转折时刻，但机器人并不需要文本，而是需要与物理世界的真实互动。为了实现这一目标，慕尼黑工业大学与 Neura Robotics 合作建立了 RoboGym，这是一个位于慕尼黑机场的 2300 平方米的训练中心，预算为 1700 万欧元。该中心的目标是收集数千小时的真实训练数据，这些数据是模拟器无法提供的。通过这种方式，机器人将能够获得更好的学习和适应能力。
机器人技术的 ChatGPT 时刻：为什么训练数据现在决定一切
📄 English Summary
Der ChatGPT-Moment der Robotik: Warum Trainingsdaten jetzt alles entscheiden
Large language models have reached a significant turning point, where vast amounts of text data suddenly endowed them with unexpected capabilities. A similar turning point is now approaching in robotics, where robots require real interaction with the physical world rather than text. To facilitate this, the Technical University of Munich (TUM) and Neura Robotics are launching RoboGym, a training center for humanoid robots covering 2,300 square meters at Munich Airport, with a budget of 17 million euros. The goal is to collect thousands of hours of real training data that simulators cannot provide. This initiative aims to enhance the learning and adaptability of robots significantly.