为什么足式机器人对具身领域如此重要？ 为什么你的足式机器人不work? 最近有位同学找峰哥咨询运控相关的问题，买了一个二开的人形，但不知 道怎么用。折腾了好久，老师给的目标是像人类一样完成爬楼梯、爬山、跑步、跳舞、翻跟头等各类动 作。该说不说，虽然具身现在工作很多，但让一个初学者上手调一个，还是很难！！！ 人形机器人和四足重强化，想象一下，机械狗能够在布满石子的路上或者山区进行工作，如果只是僵硬的 动作驱动，那估计连20m都走不出去！在面对满是石头、树木、坑洼不平的路面、各种草丛和障碍物，有 效稳定地执行动作是非常关键的。 对于开发者，真的稳定完成一套开发流程需要具备以下基础：运动学、动力学、多传感器融合、步态控 制、强化学习、模仿学习、仿真、sim2real等方向。内容众多，小白经常踩坑。 足式机器人 （四足、双足等）被业界誉为「机器人领域的下一座里程碑」，因其能像生物一样应对复杂地 形、灵活跨越障碍，不再局限于平坦环境。 资本与市场对足式机器人无比青睐，未来将在巡检、安防、救援、工业自动化等场景大放异彩。 可以说， 现在是进入足式机器人领域的最佳时机 ！相关企业对人才的需求越来越大，不惜重金求人。 然后， ...

仪器仪表Ⅲ 安培龙（301413.SZ） 增持-A(首次) 温度、压力及氧传感器夯实基础，力传感器开辟机器人应用新场景 2025 年 7 月 30 日 公司研究/深度分析 公司近一年市场表现 投资要点： | 市场数据：2025 年 | 7 月 | 29 日 | | | --- | --- | --- | --- | | 收盘价（元）： | | | 86.57 | | 总股本（亿股）： | | | 0.98 | | 流通股本（亿股）： | | | 0.53 | | 流通市值（亿元）： | | | 45.59 | | 基础数据：2025 年 3 月 | 31 日 | | | --- | --- | --- | | 每股净资产（元）： | | 12.47 | | 每股资本公积（元）： | | 7.10 | | 每股未分配利润（元）： | | 3.94 | | 资料来源：最闻 | | | 叶中正 执业登记编码：S0760522010001 电话： 邮箱：yezhongzheng@sxzq.com 冯瑞 执业登记编码：S0760524070001 邮箱：fengrui@sxzq.com 谷茜 执业登记编码：S07605 ...

Group 1 - EssilorLuxottica, the world's largest eyewear manufacturer, reported that sales of the Ray-Ban Meta smart glasses increased more than twofold, contributing to total sales of €14.02 billion (approximately $16.25 billion) in the first half of the year, a year-on-year increase of 7.3% [1] - The CEO Francesco Milleri and Deputy CEO Paul du Saillant stated that the company is leading the transformation of eyewear as the next generation computing platform, integrating AI, sensing technology, and data-rich healthcare facilities [1] - The new Oakley Meta smart glasses, launched in June, are part of the latest product line developed in collaboration with Meta, with plans for a Prada-branded smart glasses in the future [1] Group 2 - EssilorLuxottica owns several well-known brands, including Ray-Ban, Oakley, Vogue Eyewear, and Persol [2] - Meta and EssilorLuxottica have signed a long-term cooperation agreement to develop multiple generations of smart eyewear products over the next decade [2] - Meta has acquired just under 3% of EssilorLuxottica's shares, valued at approximately €3 billion ($3.5 billion), and is considering increasing its stake to around 5% [2]

Group 1: Industry Overview - The humanoid robot industry is on the brink of explosive growth, with a projected compound annual growth rate (CAGR) exceeding 60% over the next five years [4][6] - By 2030, the production capacity of humanoid robots is expected to reach 5-6 million units, with the market size surpassing 450 billion yuan [6] - In the first half of 2025, there were 87 financing events in China's humanoid robot sector, raising a total of 10.9 billion yuan, nearly doubling the financing amount from the entire year of 2024 [5] Group 2: Company Profile - Hanwei Technology - Hanwei Technology, established in 1998 and listed on the Shenzhen Stock Exchange in 2009, is a leading company in the gas sensor field in China [2] - The company is one of the few capable of mass-producing flexible tactile sensors, with a thickness of less than 0.3mm and sensitivity reaching 0.1 kPa [3] - Hanwei Technology is collaborating with various robotics companies to integrate flexible tactile sensors into robotic hands, enhancing their ability to perform complex and delicate tasks [3] Group 3: Technological Advancements - Hanwei Technology's product matrix covers multiple dimensions of perception for humanoid robots, including touch, balance, force control, and smell [4] - The company has developed various electronic skin technologies, including pressure-sensitive, piezoelectric, capacitive, magnetic, and liquid metal sensors, suitable for different body parts and functional scenarios [4] - The flexible sensor production line at Suzhou Nistad has reached full capacity, with new expansion plans underway to significantly increase production in the second half of the year [8] Group 4: Market Dynamics - The sensor market is expected to grow rapidly, driven by the humanoid robot market, with a projected CAGR of 61.6% by 2030 [7] - Force sensors are anticipated to dominate the market, with a demand scale of approximately 58.5 billion yuan by 2030, while visual sensors may reach around 20 billion yuan [7] - Hanwei Technology is actively supplying flexible electronic skin products to several humanoid robot manufacturers, indicating a strong market presence [7]

Core Insights - The CMOS image sensor (CIS) market is expected to experience strong growth in 2024, driven by the rebound in smartphone demand and needs in security, defense, aerospace, and consumer electronics. Revenue is projected to grow by 6.4% in 2024, following a 2.3% increase in 2023, with a compound annual growth rate (CAGR) of 4.4% from 2024 to 2030 [1][4]. Market Trends - The market volume is anticipated to rise from 7 billion units to 9 billion units by 2030, with mobile, security, and automotive applications being the primary growth drivers [1]. - The average selling price (ASP) remains stable above $3, supported by high-end features in mobile and automotive sectors [1]. - Wafer production is expected to grow by 8.9% in 2024, with a steady increase projected until 2030 [1]. Competitive Landscape - Chinese company Smartsens leads the market with a significant year-on-year growth of 105.7%, expanding into mobile and automotive sectors. Sony's market share has also increased by nearly 50%, while SK Hynix has reduced its focus on CIS, concentrating on memory products [4]. - Companies like ON Semiconductor, Teledyne, and STMicroelectronics have seen revenue declines due to slowdowns in industrial and medical markets, as well as reduced design orders in consumer electronics [4]. - Despite trade tensions, the CIS sector remains resilient due to the adoption of mature nodes, with domestic demand and government support bolstering production in China [4]. Technological Innovations - The CIS industry is evolving through innovations that enhance performance, integration, and sensing capabilities, including higher signal-to-noise ratios, improved low-light sensitivity, compact designs, and lower power consumption [7]. - Sony's three-layer stacked sensors are being adopted in mainstream smartphones, marking a shift towards intelligent sensing rather than just resolution [7]. - Advances in 22nm logic stacking technology aim for ultra-low power consumption and expanded computational capabilities, with FDSOI technology expected to be utilized for neuromorphic sensing [7]. Future Projections - By 2030, global CIS wafer capacity is projected to meet demand with a capacity of 638k wpm and a utilization rate of 72% [4]. - The industry is witnessing a transition towards multi-stacking architectures, with BSI (Backside Illumination) technology leading the way [8].

Core Insights - Xiaopeng Motors has been active this month, launching the new model Xiaopeng G7, which achieved over 10,000 pre-orders within 9 minutes of its release, indicating strong product and brand strength [1] - The company has shortened payment terms to suppliers to within 60 days, reflecting a robust supply chain and healthy cash flow [1] - Despite these positive developments, there are concerns regarding product line confusion and potential production cuts in the second half of the year [2] Product Line and Market Position - Xiaopeng Motors is considering product line integration due to a perceived market pessimism and internal competition among its extensive product matrix, which includes various models across different categories and sizes [2][3] - The P5 model has struggled in the market, with sales dropping from 864 units in August 2024 to just 109 units in September 2024, indicating a disconnect between its features and consumer needs [3] - The introduction of multiple derivatives of the P7 model has led to market cannibalization, with the P7+ overshadowing the P7i and affecting overall sales [4] Brand Perception and Quality Issues - The launch of the MONA M03 has led to a decline in Xiaopeng's brand positioning, which was previously considered mid-range but is now perceived as lower-end due to its association with ride-hailing services [4] - There have been reports of safety issues with the P7+ model, including steering system malfunctions, which could impact consumer trust and brand reputation [5][6] Autonomous Driving Strategy - Xiaopeng Motors has shifted its focus to a pure vision-based approach for its autonomous driving technology, moving away from laser radar due to cost considerations and the adaptability of its VLA model [7][8] - The company plans to maintain this vision-based strategy for L3 level autonomous driving while potentially reintroducing laser radar for higher levels of automation in the future [9]

三羊马(001317)最新披露的公告显示，公司于7月25日举行投资者电话会议，与光大证券等94家机构的 人士围绕三羊马智人科技（重庆）有限公司（简称"智人科技"）的相关业务展开深度交流。 此前在7月7日晚间，三羊马官方微信发布消息，旗下智人科技完成关键战略布局，成功引入冰零智能科 技 (常州) 有限公司（简称"冰零科技"）为战略投资者，为企业发展注入强劲动能。 谈及智人科技的定位，三羊马方面表示，智人科技前期以汽车传感器的销售为主、未来主要从事汽车传 感器的研发生产、业务延伸到机器人传感器的研发生产、低空传感器的研发生产。传感器业务定位主要 集中在汽车动力和底盘领域，这两个领域与安全密切相关，技术壁垒较高且国产化程度较低。其中，底 盘传感器的国产化率不超过20%，而空悬领域的传感器则基本被外资垄断。 "汽车行业准入壁垒很高，三羊马在整车物流行业深耕20年，和主机厂建立了长期稳定的合作关系，冰 零科技有很强的技术实力。双方希望通过协同效应，在西南地区的主机厂中寻求共同发展，发挥各自优 势。"三羊马方面表示。 紧接着在7月24日晚间，三羊马宣布智人科技与国内某汽车专用零部件公司签署了《产品采购合同》， 由智人科技 ...

如今，这个困局终于被打破了！ 有一种传感器，能捕捉癌细胞发出的微弱荧光，能追踪粒子对撞时的刹那闪光，甚至能"看见"宇宙边缘 的伽马射线——但它曾被德日企业垄断20年，中国90%的高端设备都得看别人脸色。 高性能硅光电倍增器（SiPM），终于实现了国产化的关键一跃！而且良品率超高，直接打破了国外巨 头长达20年的技术垄断！这意味着以后咱们用上更便宜、更先进的医疗设备、科研仪器，甚至开上更聪 明的自动驾驶汽车，都有了更坚实的"中国芯"基础！ 这个重大突破的主角是中广核技旗下的中京光电。他们成功建成了国内领先的SiPM封装生产线，并且 提前达到了一个超级厉害的目标：器件良品率超过90%！ 要知道，这种高端光电器件的生产，对洁净度要求达到"百级"，比手术室还干净，对工艺的要求也极其 苛刻，能稳定量产且良率高，标志着咱们真正掌握了核心技术，不再是"纸上谈兵"。 这条产线还拿到了欧盟的RoHS环保认证，下一步就要通过更严格的ISO体系认证，完全具备了走向国 际市场、服务全球的实力。 那么，这个SiPM到底是啥"神器"？为啥这么重要？ 你可以把它想象成一个极其敏锐的"光电转换器"，或者叫"光信号放大器"。它的核心本领是：能 ...

从融资金额来看，今年以来，亿元级融资事件的数量为12起，分别是深圳市元鼎智能创新有限公司（以 下简称"元鼎智能"）、杭州蓝芯科技有限公司、云鲸智能创新（深圳）有限公司等。其中，泳池机器人 龙头厂商元鼎智能单笔融资金额高达近10亿元，拔得头筹。 据深圳市高工产业研究有限公司数据，今年以来，已有30家移动机器人产业链企业获得融资，融资案例 多达33起（不包含人形机器人）。 千万元级融资事件的数量为19起，分别是深海智人（广州）技术有限公司（以下简称"深海智人"）、上 海库萨科技有限公司、深圳深庭纪智能科技有限公司等。其中，深圳玄源科技有限公司的单笔融资金额 将近1亿元。 从融资轮次来看，今年以来，种子轮融资有2起，天使轮融资有6轮，A轮融资有13起，B轮融资有8起， C轮融资有3轮，E轮融资有1起。 "随着人工智能、物联网、传感器等产业链技术的快速发展，移动机器人正在被广泛应用于各种场景， 如制造业、服务业、物流业等，这些领域使得移动机器人的需求正在不断增长，推动其商业化进程，带 火了这个赛道的融资，资本看好其未来的发展。"中国电子商务专家服务中心副主任郭涛在接受《证券 日报》记者采访时表示。 从融资企业的类型来 ...

Group 1 - The core viewpoint of the article emphasizes the significance of multi-sensor fusion perception (MSFP) in embodied AI, highlighting its role in enhancing perception capabilities and decision-making accuracy [5][6][66] - Embodied AI is defined as an intelligent form that utilizes physical entities as carriers to achieve autonomous decision-making and action capabilities in dynamic environments, with applications in autonomous driving and robotic clusters [6][7] - The article discusses the necessity of multi-sensor fusion due to the varying performance of different sensors under different environmental conditions, which can lead to more robust perception and accurate decision-making [7][8] Group 2 - The article outlines the limitations of current research, noting that existing surveys often focus on single tasks or fields, making it difficult for researchers in other related tasks to benefit [12][13] - It identifies challenges at the data level, model level, and application level, including data heterogeneity, temporal asynchrony, and sensor failures [12][66] - The article presents various types of sensor data, including camera data, LiDAR data, and mmWave radar data, detailing their characteristics and limitations [11][13] Group 3 - Multi-modal fusion methods are highlighted as a key area of research, aiming to integrate data from different sensors to reduce perception blind spots and achieve comprehensive environmental awareness [19][20] - The article categorizes fusion methods into point-level, voxel-level, region-level, and multi-level fusion, each with specific techniques and applications [21][29] - Multi-agent fusion methods are discussed, emphasizing the advantages of collaborative perception among multiple agents to enhance robustness and accuracy in complex environments [33][36] Group 4 - Time series fusion is identified as a critical component of MSFP systems, enhancing perception continuity and spatiotemporal consistency by integrating multi-frame data [49][51] - The article introduces query-based time series fusion methods, which have become mainstream due to the rise of transformer architectures in computer vision [53][54] - Multi-modal large language models (MM-LLM) are explored for their role in processing and integrating data from various sources, although challenges remain in their practical application [58][59] Group 5 - The article concludes by addressing the challenges faced by MSFP systems, including data quality, model fusion strategies, and real-world adaptability [76][77] - Future work is suggested to focus on developing high-quality datasets, effective fusion strategies, and adaptive algorithms to improve the performance of MSFP systems in dynamic environments [77][68]