【DT新材料】 获悉， 就在刚刚， 上纬新材 董事长， 智元机器人 联合创始人彭志辉在B站发布他在2025年的最后一个作品，介绍了新产品—— 全球首个 个人机器人，即小尺寸全身力控人形机器人产品：启元Q1。 用彭志辉的话说， 启元Q1可以成为个人探索具身智能的第一台"毕业机"。更小的尺寸，使得机器人的运动性能更高、动作更灵活，能更好更快学会动作。 如何把全尺寸人形机器人的能力压缩到不到一个书包的尺寸里？ 核心是关节微型化。 稚晖君定义启元Q1为"属于每个人的第一个个人机器人"，其站立高度约为 0.8米 ， 收纳后机身长度在 0.5米 左右， 可以折叠后装进一个30-35L左右的 双肩包里。 稚晖君还宣布，上纬新材将以 " 上纬启元 " 品牌进军个人机器人赛道，致力于将高端机器人技术转化为普通人可拥有、可操作、可创造的个人设备，这是 该公司首次对外明确机器人业务方向。 今年7月，智元恒岳入驻 上纬新材 ；11月25日，上纬新材发布公告，公司董事会选举彭志辉为董事长，任期至第四届董事会任期届满之日止。 点击阅读 ： 新材料龙头上市公司，迎90后华为"天才少年"董事长！ 为什么要做小 试错成本 是Sim2Real（ ...

" 更优雅的感知，更长程的控制。 " 作者丨 梁丙鉴 编辑丨马晓宁 编者按：12月12日， 第八届 GAIR 全球人工智能与机器人大会 于深圳正式拉开帷幕。 本次大会为期两天，由GAIR研究院与雷峰网联合主办，高文院士任指导委员会主席，杨强院士与朱晓蕊教 授任大会主席。大会共开设三个主题论坛，聚焦大模型、具身智能、算力变革、强化学习与世界模型等多 个议题，描绘AI最前沿的探索群像，折射学界与产业界共建的智能未来。 作为 AI 产学研投界标杆盛会，GAIR自2016年创办以来，始终坚守 "传承+创新" 内核，是 AI 学界思想 接力的阵地、技术交流的平台，更是中国 AI 四十年发展的精神家园。过去四年大模型驱动 AI 产业加速变 革，岁末年初 GAIR 如约而至，以高质量观点碰撞，为行业与大众呈现AI时代的前沿洞见。 在12月13日的"数据&一脑多形"专场，上海人工智能实验室青年科学家王靖博进行了以《从虚拟走向现 实，构建通用人形机器人控制与交互策略》为主题的演讲。 长期以来，人形机器人的研究是否必要一直存在着争议。演讲伊始，王靖博博士就对此做出了回应。他指 出，由人类搭建的真实生活环境，也面向人类的各种需求， ...

Core Insights - The article discusses the challenges in end-to-end robot learning, particularly focusing on the "Sim2Real" gap, which is primarily caused by the inadequacy of simulation environments to accurately replicate real-world scenarios [1][6][10]. Group 1: Challenges in Robot Simulation - Current simulation environments struggle with three main issues: insufficient realism in replicating real-world scenarios, high costs in scene asset acquisition and system configuration, and time-consuming data collection processes [1][5]. - The core obstacle is the performance drop during the Sim2Real transfer, which stems from the fundamental differences between simulated and real-world environments, such as object appearance, lighting effects, and spatial geometry [1][6]. Group 2: Existing Simulation Frameworks - Various simulation frameworks have been developed, but none meet the three critical requirements: high visual fidelity, accurate physical interaction, and efficient parallel scalability [3][6]. - Traditional simulators often compromise on either visual realism or physical accuracy, leading to ineffective training for robots [6][7]. Group 3: DISCOVERSE Framework - DISCOVERSE is an open-source simulation framework developed by Tsinghua University in collaboration with other institutions, integrating 3D Gaussian Splatting (3DGS), MuJoCo physics engine, and control interfaces into a unified architecture [5][10]. - The framework aims to bridge the Sim2Real gap by enhancing the realism of simulations through a three-layer innovation approach, focusing on accurate digital representation of real-world scenes and objects [10][12]. Group 4: Performance and Efficiency - DISCOVERSE significantly improves simulation speed, achieving rendering speeds up to 650 FPS on high-end hardware, which is three times faster than competing solutions [19][20]. - The framework supports a wide range of asset formats and robot models, enhancing compatibility and reducing the need for extensive configuration [21][22]. Group 5: Testing and Results - In comparative tests, DISCOVERSE outperformed other mainstream simulators in zero-shot transfer success rates across various tasks, demonstrating its effectiveness in real-world applications [24][27]. - The framework also enhances data collection efficiency, reducing the time required to gather demonstration data from 146 minutes to just 1.5 minutes, thus accelerating algorithm iteration [29]. Group 6: Future Implications - DISCOVERSE is positioned as a versatile robot simulation framework capable of supporting various complex tasks, with potential applications in robotics, drones, and autonomous driving sensors [30]. - The release of the framework's code and API aims to facilitate adoption by developers and enterprises, marking a significant milestone in the robotics industry [30].

Core Insights - Richtech Robotics has launched Dex, its first mobile humanoid robot designed for industrial applications, in collaboration with NVIDIA to enhance its capabilities [1][3] - Dex utilizes NVIDIA Jetson Thor technology for real-time reasoning and complex task execution, operating efficiently for a full workday on a single charge [2][5] - The robot combines insights from over 450 previous deployments, integrating autonomous mobile robot (AMR) technology with dual-armed precision to enhance operational efficiency [4][5] Technology and Development - Richtech employs NVIDIA's Isaac Sim for training Dex in diverse industrial contexts, facilitating a "Sim2Real" pipeline that accelerates deployment and improves safety [3][6] - The robot's design prioritizes mobility and dexterity, featuring a wheeled platform for stability and lower energy costs, while maintaining a four-hour battery life in mobile mode [4][5] - Richtech is launching an American robotics data initiative to collect regionally grounded data, aiming to empower the development of physical AI in the U.S. [7] Applications and Capabilities - Dex is capable of performing a variety of light to medium industrial tasks, making it a valuable asset in manufacturing and logistics sectors [8][9] - The robot's features include modular end-effectors for various tools, a four-camera vision system for navigation, and the ability to operate continuously from a static base [5][6] - Richtech invites companies to explore pilot opportunities with Dex, showcasing its capabilities at industry events [10][9] Company Overview - Richtech Robotics focuses on developing advanced robotic solutions and data infrastructure, emphasizing automation and continuous AI-driven improvement across various sectors [11]

Investment Rating - The report maintains an "Accumulate" rating for the mechanical equipment industry [4]. Core Insights - The focus of commercialization is on robotic brains, with Sim2Real potentially becoming the mainstream training method [2]. - The development of humanoid robot models is rapidly advancing, with Tesla's Optimus model demonstrating a high degree of human-like capabilities [3][29]. - The report suggests paying attention to listed companies involved in related hardware and software businesses, such as Pinming Technology [3]. Summary by Sections Section 1: Robotic Brain Development - The evolution from LLM to VLM and then to VLA models is enhancing the generalization and precision of robotic actions [1]. - VLA models are increasingly incorporating tactile inputs to improve robustness [1]. Section 2: Sim2Real Technology - Sim2Real utilizes synthetic data generation to help robots accumulate experience through diverse scenarios, linking virtual and real-world data for training [2]. - The technology involves a tri-computer setup: an AI supercomputer, a simulation computer, and a physical AI computer [2]. Section 3: Tesla's Optimus Model - Tesla's Optimus integrates AI systems from FSD and xAI's Grok model, achieving high levels of human-like interaction and physical self-awareness [3][29]. - The model's architecture allows it to process various sensory data to generate action commands directly from raw sensor inputs [33]. Section 4: Emerging Technologies and Models - The report discusses several innovative models, including RT-1, RT-2, Magma, and ViLLA, each contributing to bridging the gap between visual/textual inputs and robotic actions [14][17][22]. - The introduction of the "force-position hybrid control algorithm" by the Beijing General Artificial Intelligence Research Institute shows significant improvements in task success rates [58]. Section 5: Market Trends - The mechanical equipment industry is projected to experience varying growth rates, with a notable increase expected in the coming years [5].

Core Insights - The discussion focuses on bridging the Sim2Real gap in robotics, emphasizing the importance of simulation in training robots to operate effectively in the real world [2][4][10] Group 1: Key Participants and Context - Madison Huang, NVIDIA's head of Omniverse and physical AI marketing, made her first public appearance in a podcast discussing robotics and simulation [1][2] - The conversation featured Dr. Xie Chen, CEO of Lightwheel Intelligence, who has extensive experience in the Sim2Real field, having previously led NVIDIA's autonomous driving simulation efforts [2][9] Group 2: Challenges in Robotics - The main challenges in bridging the Sim2Real gap are identified as perception differences, physical interaction discrepancies, and scene complexity variations [4][6] - Jim Fan, NVIDIA's chief scientist, highlighted that generative AI technologies could enhance the realism of simulations, thereby reducing perception gaps [6][7] Group 3: Importance of Simulation - Madison Huang stated that robots must experience the world rather than just read data, as real-world data collection is costly and inefficient [7][9] - The need for synthetic data is emphasized, as it can provide a scalable solution to the data scarcity problem in robotics [9][10] Group 4: NVIDIA's Technological Framework - NVIDIA's approach involves a "three-computer" logic: an AI supercomputer for processing information, a simulation computer for training in virtual environments, and a physical AI computer for real-world task execution [10][11] - The simulation computer, powered by Omniverse and Isaac Sim, is crucial for developing robots' perception and interaction capabilities [11][12] Group 5: Collaboration with Lightwheel Intelligence - The partnership with Lightwheel Intelligence is highlighted as essential for NVIDIA's physical AI ecosystem, focusing on solving data bottlenecks in robotics [15][16] - Both companies share a vision for SimReady assets, which must possess real physical properties to enhance simulation accuracy [16][15] Group 6: Future Directions - The live discussion is seen as an informal introduction to NVIDIA's physical intelligence strategy, which aims to create a comprehensive ecosystem for robotics [18] - As collaboration deepens, it is expected to transform traditional robotics technology pathways [18]

Core Viewpoint - The collaboration between Guanglun Intelligent and NVIDIA aims to leverage SimReady and AI to achieve seamless migration from virtual simulation to the physical world, addressing key challenges in robot development and implementation [2][3]. Group 1: Live Broadcast Highlights - The live broadcast will focus on the technological breakthrough of Sim2Real, detailing how both companies utilize SimReady and AI to overcome challenges in robot development [2]. - Experts will share insights on the technological trends and commercialization paths in the fields of robotics and AI, drawing from their practical experiences [4]. Group 2: Collaboration Progress - Exclusive updates on the latest achievements and plans in technology research and application scenarios from the partnership between Guanglun Intelligent and NVIDIA will be disclosed [3]. Group 3: Key Speakers and Event Details - The live broadcast will feature Steve Xie, the founder and CEO of Guanglun Intelligent, and Madison Huang, Senior Director of Product Marketing at NVIDIA [6]. - The event is scheduled for October 9 at 00:00 Beijing time, which corresponds to October 8 at 09:00 Pacific time [6].

Core Viewpoint - The article discusses the ongoing debate in the field of embodied intelligence regarding the reliance on simulation efficiency versus real-world data, and the potential of world models to redefine the landscape of data utilization in this domain [4][8]. Group 1: Understanding Sim-to-Real Gap - The "Sim-to-Real gap" refers to the discrepancies between simulated environments and real-world scenarios, primarily due to incomplete simulations that fail to accurately replicate visual and physical details [8]. - Research indicates that the gap exists because simulation models do not fully capture the complexities of the real world, leading to limited generalization capabilities and a focus on specific scenarios [8][11]. - Solutions to bridge this gap involve optimizing data, including designing virtual and real data ratios and leveraging AIGC to generate diverse datasets that balance volume and authenticity [11][12]. Group 2: Data Utilization in Embodied Intelligence - There is a consensus among experts that while real data is ideal for training, the current landscape necessitates a reliance on simulation data due to the scarcity of high-quality real-world datasets in the embodied intelligence field [20][21]. - Simulation data plays a crucial role in foundational model iteration and testing, as it allows for safe and efficient algorithm testing before deploying on real machines [21][24]. - The potential of simulation in scaling reinforcement learning is highlighted, as well-constructed simulators can facilitate large-scale parallel training, enabling models to learn from scenarios that are difficult to capture in real life [24][26]. Group 3: World Models and Future Directions - The article emphasizes the significance of world models in future research, particularly in areas like autonomous driving and embodied intelligence, showcasing their potential in general visual understanding and long-term planning [30][32]. - Challenges remain in automating the generation of simulation data and ensuring the diversity and generalization of actions within simulations, which are critical for advancing the field [28][29]. - The introduction of new modalities, such as force and touch, into world models is suggested as a promising direction for future research, despite current limitations in computational resources [30][31]. Group 4: Reaction to Boston Dynamics Technology - Experts acknowledge the advanced capabilities of Boston Dynamics robots, particularly their smooth execution of complex tasks that require sophisticated motion control [33][37]. - The discussion highlights the importance of hardware and data in the field of embodied intelligence, with Boston Dynamics' approach serving as a benchmark for future developments [37][39]. - The consensus is that the seamless performance of these robots is attributed not only to hardware differences but also to superior motion control techniques that could inform future research in embodied intelligence [39][41].

Core Insights - The article discusses the development and effectiveness of SSRL (Structured Search Reinforcement Learning) in enhancing the training efficiency and stability of Search Agents using large language models (LLMs) [6][28] - SSRL demonstrates superior performance over traditional methods that rely on external search engines, achieving effective transfer from simulation to real-world applications (Sim2Real) [6][28] Group 1 - SSRL utilizes structured prompts and format rewards to effectively extract world knowledge from models, leading to improved performance across various benchmarks and reduced hallucination [2][6] - The research highlights the high costs and inefficiencies associated with current RL training methods for Search Agents, which include full-real and semi-real search approaches [7][13] - The introduction of SSRL allows for a significant increase in training efficiency, estimated at approximately 5.6 times, while maintaining a continuous increase in training rewards without collapse [31][32] Group 2 - Experiments show that models trained with SSRL outperform those relying on external engines, particularly in real-world search scenarios, indicating the importance of integrating real-world knowledge [28][31] - The article presents findings that suggest the combination of self-generated knowledge and real-world knowledge can enhance model performance, particularly through entropy-guided search strategies [34] - The integration of SSRL with TTRL (Task-Driven Reinforcement Learning) has shown to improve generalization and effectiveness, achieving up to a 67% performance increase in certain tasks [38][39]

Group 1 - The core issue in embodied intelligence is the severe shortage of real data, with most robotic models relying on less than 1% of real operational data, which limits their generalization capabilities in complex environments [5][6] - There is a debate in the industry regarding the importance of real data versus synthetic simulation data, which affects the scalability and generalization of embodied intelligence [6][7] - Some experts argue that while synthetic data has advantages in cost and scalability, it cannot fully replicate the complexities of the real world, leading to a "domain gap" that hinders model transferability [7][8] Group 2 - The need for hundreds of billions of real data points is highlighted, with current datasets only reaching the million level, presenting a significant bottleneck for the development of embodied intelligence [8] - The strategy of using synthetic data for initial training followed by fine-tuning with real data is seen as a key pathway for the cold start and scaling of embodied intelligence [8][9] - Teleoperation is emerging as a primary method for acquiring real data, especially in the early stages of embodied intelligence, where human operators provide high-quality demonstration actions for training [9][10]