与扭秧歌比，机器人有哪些"进化"？ 宇树科技介绍，为呈现本次高难度武术表演，宇树在算法、硬件与系统层面进行了系统性升级： 算法层面 ，团队升级强化学习框架，使机器人在仿真环境中自主习得复杂武术动作及器械操控技能；同时采用融合感知的定位技术，有效解决高速运动中的定位 漂移问题，实现精准导航与落点控制。 硬件层面 ，团队针 对性提升核心关节电机功率密度，优化肢体结构强度，并升级灵巧手与缓冲部件，以支撑高爆发、高冲击的动态动作 。 【DT新材料】 获悉，在2月16日晚的2026马年春晚上，宇树科技实现了多项全球首次技术突破： 机器人实现全球第一次连续花式翻桌跑酷，全球第一次 弹射空翻、空翻最大高度超3米，全球第一次单脚连续空翻、两步蹬墙后空翻，以及全球第一次Airflare大回旋七周半 等。 表演中最大难点是什么？ 宇树科技称，节目中难度最高的动作集中在三个类别，每一类都有各自的技术壁垒。 首先，物品交互动作（如棍法、双节棍、宗师剑）中， 难点在于机器人对器械状态及外部扰动的实时感知与自适应控制 。人类习武需通过反复练习形成手感， 机器人同样需要"习得"这种能力。 为此，团队通过对器械进行物理建模，在仿真环境中进行 ...

Core Viewpoint - The article highlights the technological advancements and performance of Yushu Technology's humanoid robots, particularly during their martial arts performance at the 2026 Spring Festival Gala, showcasing multiple global firsts in robotic capabilities [5][14]. Group 1: Technological Innovations - Yushu Technology's robots achieved several global firsts, including continuous parkour flips and advanced martial arts maneuvers, demonstrating significant improvements in motion performance compared to previous performances [5][6]. - The company upgraded its algorithms, hardware, and systems to enhance the robots' capabilities, including a reinforced learning framework for complex martial arts actions and improved hardware for dynamic movements [8][14]. - The core technology enabling these performances is the fully autonomous cluster control system, allowing multiple robots to perform without external positioning assistance, relying solely on onboard sensors [14][16]. Group 2: Performance Challenges - The most challenging actions in the performance included weapon interaction, environmental interaction, and extreme ground maneuvers, each presenting unique technical barriers that required advanced sensing and adaptive control [9][10]. - The robots demonstrated skills such as using weapons and executing complex movements, which required real-time perception and adjustment to external forces [10][15]. - The performance's success relied on the robots' ability to maintain precise positioning and dynamic adjustments during high-speed actions, showcasing their advanced control systems [10][16]. Group 3: Real-World Applications - The technologies developed for the performance have clear applications in real-world scenarios, such as multi-robot collaboration in industrial settings, enhancing operational efficiency and adaptability [15][16]. - Techniques for handling external forces and dynamic positioning can be applied to tasks like precision assembly, heavy lifting, and service robotics, improving robustness in various operational environments [15][16]. - The advancements in robotic capabilities during the performance are expected to facilitate the deployment of robots in complex real-world tasks, enhancing their operational efficiency and environmental adaptability [16].