证券日报网3月20日讯 ，奥比中光在接受调研者提问时表示，据了解，物理AI的目标是让智能体可以理 解真实世界的运行规律，自主地感知、理解并执行复杂操作，从而进行有效交互；而世界基础模型是理 解现实世界动态（包括其物理和空间属性）的生成式AI模型。各类AI端侧硬件通过学习，在理解现实 环境物理特性的前提下，对运动以及感知数据中的空间关系等动态进行表征和预测，实现自主交互。基 于对物理AI和世界基础模型的理解，NVIDIA已经推出多款工具类产品，用于智能驾驶、机器人训练及 工业数字孪生的开发，如NVIDIACosmos、NVIDIAOmniverse、NVIDIAIsaacSim等。 公司的3D视觉感知技术能够精准捕捉三维空间信息，结合自研算法，为各类AI智能终端赋予环境感 知、智能交互、动态导航等核心能力。作为NVIDIA全球产业数字化生态布局的合作伙伴之一，公司持 续与NVIDIAOmniverse生态深入融合并将更多视觉生态产品融入到NVIDIA平台中。如公司Gemini335、 Gemini336系列双目3D相机，已入驻NVIDIAIsaacSim机器人仿真开发平台，并与NVIDIAIsaacPercep ...

Core Insights - Cyngn was featured in an NVIDIA article that emphasizes the importance of simulation in the development of autonomous robotics [1][2] - The article highlights Cyngn's use of NVIDIA Isaac Sim for testing forklift tire dynamics on various surfaces and inclines [1][3] Company Developments - Cyngn has developed a simulation and testing environment utilizing NVIDIA Isaac Sim, which was showcased in a video of the DriveMod Tugger operating in this simulated environment [3] - The company announced the integration of high-fidelity forklift vehicle models into NVIDIA Isaac Sim, enhancing their development workflow [3][4] Technology and Product Offerings - Cyngn's DriveMod technology allows customers to implement self-driving technology without significant upfront costs or infrastructure changes [5] - The DriveMod Tugger can haul up to 12,000 lbs and automate 60 labor hours of manual material handling per week, while the DriveMod Forklift is designed for heavy loads using non-standard pallets [5]

PTC Teams with NVIDIA to Unite Design and Robotics Simulation by Connecting Onshape to NVIDIA Isaac Sim Accessibility StatementSkip Navigation BOSTON, March 17, 2026 /PRNewswire/ -- PTC(NASDAQ: PTC) today announced a new robotics design-to- simulation workflow that connects PTC's cloud-native Onshape® computer-aided design (CAD) and product data management (PDM) platformwith NVIDIA Isaac Simopen simulation framework. The workflow is being introduced at NVIDIA GTC 2026to demonstrate how teams can simulate ro ...

Core Insights - Skild AI is expanding collaborations with NVIDIA, ABB Robotics, and Teradyne Robotics to deploy its AI-powered robot brain across various industries and applications, including high-precision assembly for NVIDIA Blackwell systems with Foxconn [1][11] Company Overview - Skild AI aims to integrate AI into the physical world through its general-purpose robotics foundation model, Skild Brain, which is designed to control any robotic hardware for any task, creating a powerful data flywheel that enhances its capabilities [2][3] Technology and Innovation - The partnership with robotics OEMs will accelerate the deployment of Skild Brain, moving away from traditional task-specific programming to a model that learns directly from data, thus enabling scalability and automation [3][4] - Skild AI utilizes NVIDIA's open robotics platform and simulation frameworks to create realistic digital environments for training its robotic foundation model, allowing for extensive pretraining before real-world deployment [5][6] Collaboration and Market Impact - Collaborations with ABB Robotics and Universal Robots aim to integrate Skild Brain into their robot portfolios, facilitating automation in complex applications without the need for task-specific coding [8][9] - The integration of Skild AI's technology is expected to enhance the capabilities of industrial robots, enabling them to handle more dynamic and variable tasks, thus addressing the needs of small and medium-sized businesses and non-traditional manufacturing [10][12]

Core Viewpoint - STMicroelectronics is accelerating the global development and adoption of physical AI systems by integrating its advanced robotics portfolio with NVIDIA's robotics ecosystem, aiming to enhance the efficiency and scalability of humanoid and industrial robots [2][14]. Group 1: Integration and Collaboration - STMicroelectronics is integrating its sensors, microcontrollers, and motor control solutions with NVIDIA's Holoscan Sensor Bridge to streamline the development process for physical AI systems [14]. - The collaboration includes the integration of Leopard Imaging's stereo depth camera and high-fidelity models of ST components into NVIDIA's Isaac Sim ecosystem, facilitating faster and more accurate research and development [2][14]. - The partnership aims to simplify the connection of ST sensors and actuators to NVIDIA Jetson platforms through pre-integrated solutions, particularly benefiting humanoid robot designs [5][4]. Group 2: Simulation and Modeling - High-fidelity simulations are essential for bridging the gap between virtual training and real-world deployment, requiring substantial GPU and CPU resources [6]. - ST and NVIDIA are focused on providing accurate, hardware-calibrated models for ST components, which will enhance robot learning by reflecting real-world device behavior [7][8]. - The integration of ST's tools with NVIDIA's Isaac Sim ecosystem is expected to optimize models, thereby shortening training cycles and reducing costs associated with humanoid robotics applications [8][7].

At GTC, Cyngn Advances High-Fidelity Forklift Simulation Through FMU Integration in NVIDIA Isaac Sim Accessibility StatementSkip Navigation MOUNTAIN VIEW, Calif., March 16, 2026 /PRNewswire/ -- Cyngn (NASDAQ: CYN) announced it is integrating its high-fidelity forklift vehicle models into NVIDIA Isaac Sim, an open simulation framework for developing autonomous systems. This milestone strengthens Cyngn's autonomous forkliftprogram, including its work with Arauco, which previously pre-ordered100 autonomous for ...

Core Insights - Reply will showcase how digital twin technology and physical AI can optimize production and logistics processes at NVIDIA GTC 2026, highlighting the integration of digital and physical worlds in industrial environments [1][1][1] Group 1: Event Details - The NVIDIA GTC conference will take place from March 16 to 19, 2026, in San Jose, California, with over 30,000 participants expected from more than 190 countries [1][1][1] - Reply will present two use cases: self-learning edge AI in manufacturing and logistics, and an intelligent robot coordination system for the Otto Group [1][1][1] Group 2: Technological Solutions - On March 17, Reply will introduce "The AI Fast Lane for the Industrial Edge powered by NVIDIA on AWS," which optimizes AI models on connected edge devices, processing sensor data in real time and ensuring model quality through a human-in-the-loop approach [1][1][1] - Reply and Google will jointly present a solution for intelligent robot coordination using NVIDIA Isaac Sim on Google Cloud, enabling the creation of precise digital twins for logistics environments and accelerating validation processes [1][1][1] Group 3: Case Study - Otto Group - The Otto Group will demonstrate a digital twin that replicates its warehouse and robotic systems, showcasing centralized fleet coordination and optimized processes during peak periods [1][1][1] - The project will be presented on March 17, focusing on leveraging physical AI to simulate and orchestrate robotic fleets for retail fulfillment centers [1][1][1] Group 4: Company Overview - Reply specializes in designing and implementing solutions based on new communication channels and digital media, supporting key industrial sectors such as telecom, media, banking, and public administration [1][1][1]

Group 1: Digital Twins and Physical AI - Reply is showcasing the integration of digital simulation and physical automation at NVIDIA GTC, highlighting projects for the Otto Group and the industrial edge domain [1] - The focus is on two use cases: self-learning edge AI in manufacturing and logistics, and an intelligent robot coordination system for the Otto Group [1] - The application "The AI Fast Lane for the Industrial Edge powered by NVIDIA on AWS" optimizes AI models on autonomous edge devices, processing sensor data in real time [1] Group 2: Collaboration and Technology - Reply and Google are jointly presenting a solution for intelligent robot coordination using cloud-based robot simulation, enabling the creation of precise digital twins without local workstations [1] - The Otto Group will demonstrate intelligent robot coordination using a digital twin that replicates the warehouse and robotic systems, leading to centralized fleet coordination and optimized processes [1] - The integration of NVIDIA Omniverse with Isaac Sim allows continuous improvements and efficiency gains to be validated using synthetic data [1] Group 3: Financial Performance - Reply Group reported a consolidated turnover of €2,483.6 million for 2025, an increase of 8.0% from €2,300.5 million in 2024 [2] - Consolidated EBITDA for the same period was €467.6 million, indicating positive financial indicators [2]

Core Insights - Cyngn is advancing its collaboration with NVIDIA by developing a simulation environment using NVIDIA Isaac Sim to enhance the commercial deployment of its autonomous vehicle solutions [1][2] Group 1: Collaboration and Technology Development - The simulation environment allows Cyngn to run its autonomy and fleet management software in a high-fidelity digital warehouse, facilitating faster validation and refinement of capabilities [1][3] - This environment supports larger simulated fleets and more complex operational scenarios, enabling Cyngn to accelerate quality assurance cycles and evaluate new features earlier in development [3][4] Group 2: Strategic Impact - The use of NVIDIA Isaac Sim is seen as a critical factor in bringing new autonomous products to market, allowing for faster validation of new use cases and reducing development risks [4] - Cyngn is contributing a detailed industrial-vehicle dynamics model to the Isaac Sim framework, which aims to improve simulation accuracy for real-world performance [4] Group 3: Future Applications - Cyngn plans to utilize the new simulation environment for development, customer demonstrations, and early-stage training workflows, while also exploring opportunities with existing partners [5] - The company’s DriveMod technology enables customers to integrate self-driving technology into their operations without significant upfront costs [7]

Core Insights - Zhiyuan Robotics launched the open-source simulation platform Genie Sim 3.0 at CES, powered by a large language model [2][4] - Genie Sim 3.0 integrates 3D reconstruction and visual generation, enabling the creation of high-fidelity digital twin environments [2][4] - Developers can generate thousands of training and testing scenarios within minutes by inputting natural language commands [2][4] Data and Features - The company also released a comprehensive simulation dataset containing over ten thousand hours of real robot operation scenarios, covering more than 200 tasks [2][4] - The dataset includes multi-sensor information such as RGB-D, stereo vision, and full-body joint states, addressing various generalization dimensions like background, layout, lighting, and noise [2][4]