还记得波士顿动力 Atlas的跑酷视频吗？当年那段画面，让全世界真切感受到人形机器人运动能力的大跨步突 破。早年也有机器人跳舞的演示，技术爱好者会盯着看关节怎么动、平衡稳不稳 ， 那时候的行业，总爱比谁 能做出更酷炫的极限动作。 但短短半年，风向彻底变了。现在打开机器人企业的新品视频，跑酷、跳舞少见了，取而代之的全是 "叠衣 服" 的操作。 Figure 03用五指手试着叠毛巾，边角偶尔卷起来也没停下；Weave Robotics的半自动叠衣视频加了2倍快 进，看着利落， 实际上 藏着真实速度偏慢的问题 。 机器人从疯狂炫技再到着手做家务， 说到底是行业 当中的概念炒作少了。企业开始重新思考逐渐开始 触碰 市场的真实需求。 谷歌 ALOHA的挂衣演示没剪帧，动作慢悠悠的，偶尔还对不齐衣架，反倒因为真实圈了不少粉；Dyna Robotics更直接，让机器人连续18小时叠餐巾，机械臂反复起落，透 着一 股死磕一个任务的执着劲儿。 ▍ 为什么偏偏 要学习 叠衣服？ 机器人企业扎堆做叠衣，核心是技术和需求对上了 。 十年前，机器人叠衣服还是实验室里的稀罕事。 2010 年由 Willow Garage 公司推出来的 ...

Core Insights - The article discusses the advancements in embodied intelligence, highlighting its integration with artificial intelligence to enhance productivity across various sectors, including industrial manufacturing and service industries [1][3]. Group 1: Embodied Intelligence Applications - Boston Dynamics' Atlas robot is capable of precise part sorting in industrial settings [1]. - CASBOT W1 from Zhongke Huiling can quickly switch between different product adaptations within one hour [1]. - The STAR1 humanoid robot from Xingdong Jiyuan can perform complex tasks like screw fastening and tool operation in automotive assembly [1]. - Linkerbot's piano robot surpasses human dexterity in playing the piano with its bionic hands [1]. - Galbot supports full-service operations in the "Galaxy Space Capsule" pop-up store, serving an average of 2000 visitors daily [1]. - The "Jueying X30" quadruped robot from Yundongchu Technology conducts detailed inspections in substations, even in harsh environments [1]. Group 2: Competition Overview - The 2025 Second Zhongguancun Embodied Intelligence Robot Application Competition aims to push robots to effectively perform tasks in real-world scenarios, covering four core areas: industrial manufacturing, commercial services, home services, and emergency response [3][4]. - The competition will assess robots' practical capabilities through two modes: autonomous completion and remote operation [6]. Group 3: Competition Structure - The competition consists of three phases: registration, preliminary rounds, and finals, with the registration phase starting on July 15 [4][8]. - The preliminary rounds are scheduled for November 4-5, divided into East Sheng and Wenquan competition zones, each responsible for specific tasks [8][9]. - The finals will take place on November 17-18 at the Zhongguancun National Independent Innovation Demonstration Zone, with a prize ceremony on November 19 [9]. Group 4: Task Categories and Requirements - The competition features two main categories: autonomous completion and remote operation, with a total of six sub-events and 15 individual tasks designed to simulate real work scenarios [6][10]. - Each task in the autonomous category must be completed without human intervention, focusing on core capabilities like navigation and precise operation [10][13]. - The remote operation category allows for human intervention but relies on the robot's sensors for environmental perception [17][19]. Group 5: Strategy Recommendations - Teams are advised to focus on tasks that align with their technical strengths and avoid overly challenging tasks that exceed their robots' capabilities [23][24]. - Emphasis should be placed on optimizing human-robot interaction efficiency and ensuring safety measures are in place during the competition [24][27]. Group 6: Competition Support and Incentives - The competition offers a total prize pool of 2 million yuan, with specific awards for each task and additional incentives for innovative solutions [25][27]. - Winning teams may receive support in terms of funding, talent, and resources if they establish operations in Haidian [27][29].

Core Insights - The article discusses the evolution of human-machine communication, emphasizing the need for humans to prepare for a future where machines can decode human emotions and expressions [1][6] - The book "The Rules of Human-Machine Communication" by John Maeda explores the philosophical implications of this evolution, highlighting the dual background of the author as both an artist and a programmer [3][4] Group 1: Human-Machine Interaction - The book outlines the characteristics of intelligent machines, noting their ability to perform repetitive tasks tirelessly and the exponential growth of data and computing power that enhances their capabilities [5] - It emphasizes the distinction between machine intelligence and human emotions, stating that machines currently lack the ability to understand human feelings and subconscious [5][6] Group 2: Communication Paradigms - Human communication is categorized into three dimensions: syntax (information transfer), semantics (intent understanding), and pragmatics (emotional resonance), with AI currently excelling in the first two [6] - Research indicates that when AI dialogue systems achieve a 63% match in pragmatic communication, human trust in these systems increases significantly [6] Group 3: Ethical Considerations - The article raises concerns about algorithmic bias, illustrating how even a small percentage of bias in training data can lead to significant systemic discrimination after multiple iterations [6] - It questions the need for a new form of "mechanical empathy" as AI systems begin to mimic human language behaviors, suggesting that this transformation will redefine human identity [6]

困难在于，马斯克想造人，而不是机器。 文丨 贺乾明 李梓楠 编辑丨贺乾明 黄俊杰 "就算我明天被外星人绑架，特斯拉也会解决自动驾驶问题。但我不知道特斯拉怎么在 Optimus 上赢。" 今 年 4 月的财报会上，马斯克给出他该拿 500 亿美元薪酬的一个理由。 一位接触过特斯拉团队的人告诉我们，在马斯克说这番话前不久，他就要求人形机器人团队在今年内拿出 有意义的突破——让机器人能上产线完成部分工作，如果达不到就进一步裁员，"团队承受的压力越来越 大"。 我们得知，Optimus 人形机器人目前已经生产数百台，但还不到可以上产线工作的状态，目前只能用来采 集数据、测试不同的设计方案。 这个 2021 年开始推进的人形机器人项目与特斯拉开发其他产品的逻辑截然相反。造车，特斯拉从高利润 的电动跑车 Roadster 起步，逐步提升生产效率，造出便宜的 Model 3。自动驾驶也是从低到高，边卖边提 升。 最大的困难：造人，而不是造人形机械装置 根据《马斯克传》，2021 年初，马斯克在特斯拉高管会上播放了波士顿动力人形机器人 Atlas 的视频，画 面中人形机器人可以像人一样，灵活完成高难度动作。他告诉参会的高管， ...