Robotics

Search documents
为软体机器人穿上“铠甲”!受虾类启发,中国团队造出刚柔并济的机械外骨骼
机器人大讲堂· 2025-08-23 04:07
想象一下,机器臂能像章鱼触手一样灵活缠绕,又能瞬间变得如钢铁般坚硬,稳稳提起重物。这看似是软体机 器人领域一个遥不可及的梦想。而现在,这一愿景正逐渐照进现实。 来自军事科学院国防科技创新研究院的软体机器人外骨骼文章登上《 Science Advances》杂志封面 如今 软体机器人是机器人领域一个迅速发展的分支 , 其核心特点在于利用柔性材料构建机器人的主体和执 行机构,使之能够实现连续、大范围的形变运动。但是,一个致命的 "软肋"长期阻碍着它们走出实验室。材 料柔软带来更好的顺应性和安全性,却也导致其结构刚度低、输出力有限,难以完成诸如重物抓取、大力操控 等任务。为了让它"硬起来",研究人员们尝试了种种方案:可变刚度材料、颗粒阻塞、液器压力锁紧……但 它们往往在刚度变化范围、响应速度或系统复杂性上存在局限。 正是在这一技术背景下,来自 中国军事科学院国防科技创新研究院的研究团队 从自然界的虾蛄等生物中获得 灵感,这些生物的外骨骼精巧地融合了刚性与柔性区域,使其在保持身体灵活的同时又足够坚固以承受巨大的 冲击力。 仿照这种 "刚性-柔性耦合"的设计原理,他们成功开发出一种基。 近日,该成果以 Origami ...
China's push for global AI dominance
NBC News· 2025-08-23 02:18
If there is a race for global domination and artificial intelligence, this may well be the starting block for China. This is Hjo, a place rich in ancient history, now emerging as China's Silicon Valley. Home to Alibaba and Deep Seek, and coming into its own as one of the world's leading tech incubators.>> I think a good idea should fulfill the three criteria. It's something that you want to do, something you're really good at, and something the society needs. >> Jang Jia is a tech entrepreneur and angel inv ...
Trend-setting Tech Feast! 33rd Guangzhou Fair Meets International Invention Exhibition
2 1 Shi Ji Jing Ji Bao Dao· 2025-08-23 01:56
Group 1 - The 33rd Guangzhou Fair (GZF) commenced on August 22, covering approximately 180,000 square meters, which is over a 50% increase in scale compared to the previous edition, attracting more than 3,200 business associations and enterprises as buyers [1] - This year's fair emphasizes the development of new quality productive forces, focusing on emerging industries such as artificial intelligence, robotics, intelligent connected vehicles, new energy vehicles, low-altitude economy, and aerospace [2] - The newly established Guangzhou New Quality Productive Forces Exhibition Zone features advanced intelligent manufacturing technology and innovative products, including a Robot Performance Area showcasing over 130 robots for an interactive experience [2] Group 2 - Concurrently with the GZF, the 11th International Invention Exhibition and the Belt and Road & BRICS Skills Development and Technological Innovation Competition are taking place, with over 300 guests from more than 30 countries and international organizations attending [3] - More than 2,000 inventions from China and around the world are on display at the event, highlighting global innovation [3]
苏州肯思尼机器人有限公司成立 注册资本200万人民币
Sou Hu Cai Jing· 2025-08-22 23:23
天眼查App显示,近日,苏州肯思尼机器人有限公司成立,法定代表人为叶子青,注册资本200万人民 币,经营范围为一般项目:智能机器人销售;智能机器人的研发;技术服务、技术开发、技术咨询、技 术交流、技术转让、技术推广;人工智能硬件销售;人工智能公共服务平台技术咨询服务;人工智能公 共数据平台;人工智能理论与算法软件开发;人工智能通用应用系统;信息系统集成服务;人工智能应 用软件开发;人工智能基础软件开发;人工智能行业应用系统集成服务;智能控制系统集成;集成电路 设计;软件开发;信息技术咨询服务;翻译服务;计算机软硬件及辅助设备批发;计算机软硬件及辅助 设备零售;市场调查(不含涉外调查);互联网数据服务;会议及展览服务;货物进出口;技术进出口 (除依法须经批准的项目外,凭营业执照依法自主开展经营活动)。 ...
Getting In The Ring With A Chinese Robot Boxer
Bloomberg Technology· 2025-08-22 17:01
It's very likely that human robots are going to be robots that we can deploy into the world relatively easily. From labs in Shenzhen to workshops in Silicon Valley. Humanoids are stepping out of science fiction and into our reality.The race is on to dominate the next era in robotics, with companies in Asia and the US vying to take the lead. China is seen as a frontrunner, with Morgan Stanley saying the country will have the highest number of humanoids in use worldwide by 2050. It's the result of many years ...
From Asimo to Agibot: Asia's humanoid evolution
Bloomberg Technology· 2025-08-22 14:02
It was 1999 as people were going crazy over the Y2K bug. Honda unveiled a piece of tech that gave the world a glimpse of what the new millennium could look like as the birthplace of Gundam Volta's five and Astro Boy, Japan's humanoid robots have stood as symbols of a tech driven future. A quarter of a century since ASIMO robots have become pervasive in modern Japanese society, driven by need.Faced with a severe labor shortage and one of the world's fastest aging populations, businesses are investing in robo ...
国泰海通|机械:Figure人形机器人实现无遮挡行走,能力边界持续突破
国泰海通证券研究· 2025-08-22 09:08
Core Viewpoint - The humanoid robot industry is experiencing rapid advancements, particularly in the area of "human-like autonomous decision-making," driven by technological deepening and practical applications [1][2][3] Group 1: Technological Breakthroughs - Figure's humanoid robot has demonstrated significant progress in walking technology, showcasing the ability to navigate complex environments without visual input [1] - The robot's walking stability has improved through reinforcement learning, achieving superhuman performance in certain areas [1] - The introduction of the new Helix walking controller and the ability to autonomously fold clothes highlight the robot's advanced skill set and adaptability [1][2] Group 2: Technical Evolution - The evolution of humanoid robots is moving from motion control and perception interaction to multi-scenario adaptability, requiring high precision in environmental perception and force control [2] - Key technologies such as tactile feedback, algorithms, and data accumulation are crucial for expanding the robot's skill set, enabling dynamic decision-making and precise force control [2] Group 3: Industry Dynamics - The industry is expected to maintain a dual-driven momentum of "technological deepening" and "scene implementation," with a collaborative upgrade path involving brain, algorithms, data, and components [3] - Humanoid robots are gradually penetrating from industrial applications into household scenarios, indicating a potential expansion of their capabilities [3]
Science Robotics 通过人机交互强化学习进行精确而灵巧的机器人操作
机器人圈· 2025-08-22 09:02
机器人操作仍然是机器人技术中最困难的挑战之一,其方法范围从基于经典模型的控制到现代模仿学习。尽管这 些方法已经取得了实质性进展,但它们通常需要大量的手动设计,在性能方面存在困难,并且需要大规模数据收 集。这些限制阻碍了它们在实际世界中的大规模部署,其中可靠性、速度和稳健性至关重要。 强化学习 (RL) 提供了一种强大的替代方案,它使机器人能够通过交互自主获得复杂的作技能 。然而,由于样品效率和安全性问 题,在现实世界中充分发挥 RL 的潜力仍然具有挑战性。 强化学习 (RL) 是一种很有前途的方法,可以自主获取复杂而灵巧的机器人技能。通过反复试验学习,原则 上,有效的 RL 方法应该能够获得针对部署任务的特定物理特征量身定制的高度熟练技能。这可能会带来不仅超 过手工设计控制器的性能,而且超越人类远程作的性能。然而,由于样本复杂性、假设(例如,准确的奖励函 数)和优化稳定性等问题,在现实环境中实现这一承诺一直具有挑战性。RL 方法对于模拟训练和现有大型真实世 界数据集的训练非常有效,目的是泛化 。它们还与手工设计的功能或表示一起使用,用于狭隘的定制任务。然 而,开发通用的、基于视觉的方法仍然具有挑战性,这些方法 ...