谁能想到，给草坪 " 理发 " 的 割草 机器 人 ，如今成了海外市场的香饽饽？ 海关总署数据显示， 2024 年我国割草机（海关编码 84331100 ）出口额达到 145 亿元， 2018-2024 年复 合增速达到 17.4% ，远高于全球割草机近 10 年约 2% 的复合增长。 2025 年 1-7 月我国割草机出口额达到 132 亿元，同比增长 56.6% 。 2025 年上半年，全球割草机器人销量同比暴涨 327% ，狂卖 234 万台，其中 主要由 中国企业推出的 " 无 边界款 " 由 2024 年 35% 提升至 65% ， 全球销售出 152 万台， 是增长的主要动力。 从欧美家庭后院到高尔夫球场，越来越多的草坪上，再也看不到弯腰推 割草 机器 设备 的 工人 ，取而代之的 是能自己规划路线、精准割草的智能机器人。 这场 " 草坪革命 " 背后，藏着一个百亿美元市场的新机遇，而中国企业正凭技术 与产品力 优势，把传统巨头 甩在身后。 ▍ 技术突破 正 改写行业规则 1995 年，割草机器人就已诞生，但那时的产品堪称 " 麻烦制造者 " ， 想用它，得先在草坪下埋一圈边界 线，改次草坪形状就 ...

Group 1 - The article highlights the rapid development and commercialization of humanoid robots, with several companies securing significant orders and launching innovative products in various sectors [1][2][4] - Notable contracts include a 250 million yuan procurement agreement by UBTECH for humanoid robot products, and strategic partnerships aimed at deploying over 1,000 humanoid robots in logistics and manufacturing [1][2] - The upcoming second Zhongguancun Embodied Intelligence Robot Application Competition aims to showcase the practical applications of humanoid robots, providing a platform for testing and evaluating their capabilities in real-world scenarios [2][4] Group 2 - The competition will focus on various application scenarios, including home services, commercial services, industrial manufacturing, and safety disposal, emphasizing the practical skills of participating robots [9][10] - It will feature three main tracks: Embodied Intelligence Model Capability Challenge, Embodied Intelligence Scene Application, and Embodied Intelligence Academic Frontiers and Industrial Ecology, each with specific goals and evaluation criteria [10][19] - The event aims to foster innovation, enhance collaboration between academia and industry, and accelerate the commercialization of advanced robotic technologies [32][33] Group 3 - The competition will offer substantial rewards, including cash prizes and opportunities for internships and project collaborations, to attract top teams from around the world [17][27] - Participants will have the chance to showcase their latest technological advancements and engage with industry experts, potentially leading to new partnerships and research initiatives [30][32] - The event is positioned as a significant opportunity for teams to demonstrate their capabilities and contribute to the evolution of the humanoid robotics industry [30][33]

Core Insights - The article highlights the rapid growth of the embodied intelligence robotics industry in Zhongguancun, particularly in Haidian District, which has established a comprehensive industrial chain with 264 innovative companies and 24 humanoid robot manufacturers, supported by a talent pool of 19,000 professionals [1][3][4] Industry Development - The successful hosting of the 2024 Zhongguancun Bionic Robot Competition has injected strong momentum into the industry, attracting 101 teams nationwide and showcasing breakthroughs in biobionics, mechanical electronics, and artificial intelligence [3][4] - The upcoming 2025 competition will further upgrade its scope to cover industrial, household, and security applications, solidifying Zhongguancun's position as a global hub for embodied intelligence technology exchange [3][4] Talent and Technology Integration - The competition serves as a platform for connecting talent, technology, products, and capital, accelerating the transition from laboratory results to industrial applications [4][5] - Local teams from Haidian won 20 awards in the 2024 competition, demonstrating the region's innovative ecosystem and its ability to attract and retain talent [3][4] Commercial Success Stories - The company CASBOT, which won multiple awards in the 2024 competition, successfully secured nearly 100 million yuan in angel financing and launched its first industrial-grade product, "CASBOT 01," which is designed for hazardous tasks [5][6] - Another team, BQR3, developed a four-legged robot that successfully operated in complex environments, leading to significant orders from automotive manufacturers and partnerships for security applications [8][6] Upcoming Competition Structure - The 2025 Zhongguancun Embodied Intelligence Robot Application Competition will feature three main tracks: model capability challenges, scene application competitions, and academic frontiers and industrial ecology contests, aimed at addressing industry needs and fostering rapid growth for participating teams [10][12] - The competition will provide a total prize pool of 2 million yuan and various support policies for winning teams, facilitating connections with key industry players and investment institutions [14][15]

Core Viewpoint - The launch of the X7 series by Daka Robotics marks a significant shift in the humanoid robot industry, with a competitive pricing strategy that positions it as a strong contender in the market, offering products at a fraction of the cost of existing alternatives [1][3][6]. Group 1: Product Launch and Pricing Strategy - Daka Robotics has introduced the X7 series humanoid robots, with an early bird price starting at 69,800 yuan, which is only one-fifth of similar products in the industry [1]. - The X7 series has achieved remarkable recognition, securing a top-three finish among 280 global teams at the recent World Humanoid Robot Competition, demonstrating its advanced capabilities [1][3]. - The pricing strategy is based on ensuring users can recover costs within 12 months, which is a reverse approach compared to the industry norm of product-first, pricing-later [8]. Group 2: Technological Advancements - The X7 series is equipped with the XBrain 2.0 commercial model, which utilizes a unique multi-layer perception model architecture, differing from the mainstream Vision-Language-Action (VLA) framework [9][11]. - This model allows for efficient task execution and adaptability to new scenarios with minimal data updates, enhancing operational stability [11]. - Daka Robotics has focused on optimizing every aspect of the robot's design, from power systems to sensor selection, to achieve cost-effective production without compromising performance [8][9]. Group 3: Market Position and Future Outlook - Daka Robotics has received over 1,000 intention orders for the X7 series, indicating strong market interest and competitive positioning against existing products [12]. - The humanoid robot industry is experiencing a dichotomy, with significant project contracts being awarded while many products still show zero sales, highlighting the need for practical applications and market acceptance [15][16]. - The global humanoid robot market is projected to grow significantly, with estimates of reaching $5.3 billion by 2025, suggesting a robust future for companies that can effectively integrate their products into commercial scenarios [18].

会 议将组织参会者 走进一线企业： 「 宇树科技 」「 微分智飞 」 。 我们诚挚邀请国内外学界同仁、科研工作者及行业工程师踊跃报名， 共筑新知。 2025年10月18日-19日， 第三届自主机器人技术研讨会 （ARTS 2025）将在中 国· 浙江大学 （玉泉校区）盛大举办 。 在常规学术议程基础之上，ARTS 2025 进一步推出 ARTS奖学金 、 学术辩论赛 、 学术吐槽大会 （脱口秀） 及 企业参观。 旨在打破传统会议单向输出的模 式，让思想的碰撞不止于论文宣讲，构建产学融合与坦诚对话的立体化交流场景。 了解会议详情，扫码加入 【 ARTS 2025 交流群】 ARTS 2025 组织机构 主办单位 ： 中国自动化学会 承办单位 ： 浙江大学控制科学与工程学院 上海交通大学自动化与感知学院 协办单位 ： 深蓝学院 ARTS 2025 会议议程 | | 09:00-09:10 开幕式 | | | --- | --- | --- | | | 09:10-09:20 | 处 视频 | | | 09:20-09:50 | 报告: 柔性人工肌肉技术及其机器人应用 | | | | 嘉宾: 赵慧蝉,清华大学长聘副 ...

Group 1 - Neuralink demonstrated a brain-machine interface allowing ALS patient Nick Wray to control a robotic arm for daily tasks, marking a significant advancement in neurotechnology [3][4] - Wray, who has undergone the N1 chip implantation, expressed excitement about regaining independence in activities like eating and using a microwave [3] - This demonstration is part of Neuralink's CONVOY research project, with Wray being the eighth patient to receive the brain implant [3] Group 2 - Giant Crab Intelligent, a startup focused on high-performance robotic joints, completed nearly 100 million yuan in Series A financing led by Ecovacs Investment [6] - The company aims to break foreign technology barriers and develop competitive high-end core components for robotics, having established a modern manufacturing base of 8,000 square meters [6] Group 3 - MIT and Toyota Research Institute developed a "controllable scene generation" method to create realistic virtual training environments for robots [10] - This method utilizes a dataset of over 44 million 3D rooms to simulate real-world interactions, enhancing the training of robotic systems [10] Group 4 - Neptune Robotics secured $52 million in Series B funding, led by Granite Asia, to develop underwater robots for ship hull cleaning [13] - The robots can clean large vessels efficiently, completing tasks 3 to 5 times faster than human divers, and can operate in various challenging conditions [13] Group 5 - A student team from Guizhou University developed a high-performance earthquake exploration and rescue robot, integrating edge computing and soft robotic arms [16] - The robot has been tested in real rescue applications and can provide critical data support for rescue teams in disaster zones [16]

在 一些 电影 片段 中，我们常常看到机器人能轻松理解 "帮我拿杯咖啡"或"把衣服晾到阳台"这样的指令， 并精准完成任务。然而，在现实世界中，让机器人具备这种高级别的理解和执行能力，一直是科研人员面临的 巨大挑战。 这背后，其实是机器人具身导航技术的现实困境。前主流的机器人导航方法，主要有两类，但都有明显的短 板。 指令跟踪导航 ： 机器人需要遵循极其详细的、一步步的指令，例如 "左转，出门，直行三米"。 但这种方式 指令依赖度 过 高，要求人类提供超出自然交互习惯的细节化引导 ； 预定义物体导航 ： 机器人被要求找到某个类别（如 "椅子"）的任意一个实例。这忽略了具体情境，比如 在"我想喝咖啡"这个任务中，找到一台"咖啡机"远比找到任意一把"椅子"更重要。 这两种方式都存在明显的 "语义鸿沟 "。 机器人 无法理解人类高级指令背后真正的意图，也无法处理 "茶室 左边桌子上的水果"这样包含复杂空间关系的任务。现实世界的人类指令是开放、抽象且充满上下文关联的， 这使得现有的导航机器人在面对真实、复杂的家居或办公环境 的 长视域导航任务中成功率极低 。 ▍ NavA ³：给机器人装 "分层导航系统" 针对以上困境 ...

机器人大讲堂获悉， 10 月 11 日，枢途科技正式对外披露已完成数千万元天使轮融资，本轮由东方富海及 兼固资本联合领投，值得一提的是， 2024 年枢途科技就拿到过数百万融资，投资方为奇绩创坛。 众所周知，数据是解决具身智能落地的最后一公里，而具身智能数据却存在采集成本高、规模化数量少、场景 多样性稀缺等问题。目前行业内主流的解决方案是采用人工数据采集方式获取高质量训练数据，仅 2025 上 半年，上海、北京及合肥等地就已构建起多个大型数采中心。 其中上海张江机器人谷，是全球唯二、全国唯一的规模化机器人数采中心，累计投放近百台数采机器人；北京 石景山首钢园人形机器人数据训练中心，累计部署超过 108 台机器人；而正在构建的合肥市具身智能机器人 数据采集预训练场预计年底将实现超 100 台机器人同时在 30 个以上细分场景开展数据采集和协同作业。 目前单台数采设备从几万到十几万不等 ，虽然行业内有部分研究机构推出低成本硬件数据采集方案，实现对 人类动作的 1:1 复刻，但在通用性与数据采集效率上，依然无法满足现有的行业发展需要。 因此谁能解决 " 低成本、高通用、多模态 " 具身智能数据问题，谁便拿到具身智能 ...

机器人大讲堂获悉， 10 月 11 日，枢途科技正式对外披露已完成数千万元天使轮融资，本轮由东方富海及 兼固资本联合领投，值得一提的是， 2024 年枢途科技就拿到过数百万融资，投资方为奇绩创坛。 众所周知，数据是解决具身智能落地的最后一公里，而具身智能数据却存在采集成本高、规模化数量少、场景 多样性稀缺等问题。目前行业内主流的解决方案是采用人工数据采集方式获取高质量训练数据，仅 2025 上 半年，上海、北京及合肥等地就已构建起多个大型数采中心。 其中上海张江机器人谷，是全球唯二、全国唯一的规模化机器人数采中心，累计投放近百台数采机器人；北京 石景山首钢园人形机器人数据训练中心，累计部署超过 108 台机器人；而正在构建的合肥市具身智能机器人 数据采集预训练场预计年底将实现超 100 台机器人同时在 30 个以上细分场景开展数据采集和协同作业。 目前单台数采设备从几万到十几万不等 ，虽然行业内有部分研究机构推出低成本硬件数据采集方案，实现对 人类动作的 1:1 复刻，但在通用性与数据采集效率上，依然无法满足现有的行业发展需要。 因此谁能解决 " 低成本、高通用、多模态 " 具身智能数据问题，谁便拿到具身智能 ...

Core Insights - The article discusses the significance of the Conference on Robot Learning (CoRL) and highlights the achievement of a Chinese research team winning the Outstanding Paper Award for their work on UniFP, a unified control algorithm for legged robots [1][3]. Group 1: Conference Overview - CoRL is a leading academic conference in AI and robotics, showcasing cutting-edge research in robot learning [1]. - In 2025, CoRL received nearly 1,000 submissions, with 264 papers accepted after rigorous review [1]. Group 2: UniFP Algorithm - UniFP (Unified Force and Position Control Policy) is the first algorithm in legged robotics to unify force and position control within a single framework, overcoming traditional limitations [3][4]. - The algorithm is based on biomechanical impedance control principles, allowing robots to respond to environmental forces similarly to human muscle perception [3][4]. Group 3: Control Framework - The UniFP framework consists of three core modules: observation encoder, state estimator, and actuator, forming a complete control loop of perception, decision-making, and execution [7]. - The observation encoder processes historical state information, while the state estimator infers unmeasurable key states, enabling "sensorless force perception" [7][8]. Group 4: Performance Validation - The research team validated UniFP in various simulated contact scenarios and later on the Unitree B2-Z1 quadruped robot, demonstrating impressive multi-functional capabilities [8][10]. - In experiments, UniFP showed precise force control, adaptive force tracking, and compliant impedance control, outperforming traditional methods [10][11]. Group 5: Imitation Learning - UniFP's integration with imitation learning significantly enhances the robot's learning efficiency in contact-intensive tasks, achieving a success rate improvement of approximately 39.5% over traditional methods [11][13]. - The research demonstrated UniFP's versatility across different robot forms and tasks, confirming its generalizability [13][14].