邊緣運算與機器人 精選議程

Humanoid robotics hardware is seeing a major acceleration in progress. It's reasonable to expect commercially viable humanoid hardware to exist within three years. The question is, how do we build the AI that can make these robots helpful? Humanoids have high dimensional action spaces and high dimensional multimodal observation space, posing challenges to existing paradigms. We'll explore these challenges in more depth and highlight some promising paths forward.

We'll explore the essential role of AI in enhancing workplace safety. We'll present a comprehensive overview of developing vision-based AI solutions at the edge to create safer and smarter industrial environments, significantly reducing the risk of incidents.

Developers are building more intelligent, responsive, and efficient humanoid robots capable of transforming industries such as manufacturing, healthcare, and retail. NVIDIA is building Isaac GR00T — a platform for developing humanoid robots. Isaac GR00T has four parts: robot foundation models for cognition and control, simulation frameworks built on NVIDIA Omniverse and Cosmos, data pipelines for generating synthetic data and environments, and finally — a supercomputer in the robot Jetson AGX Thor. This engaging session will cover the latest advancements in each of the four pillars of the platform built by NVIDIA Research and Engineering through concrete use cases and examples.

Google deploys AI models onto humanoid robots that perform tasks in the real world. Training robots to operate in diverse, complex environments (e.g., any possible kitchen) requires a “simulation-first” approach, starting with generating large sets of synthetic data for training and simulation for validating robot behavior at scale. Learn how Google improved its simulation performance by integrating core NVIDIA technologies and libraries, including Warp, into critical parts of its humanoid simulation pipeline.

Artificial intelligence is unlocking new use cases in industrial, robotics, and medical, in which safety and security are a key concern. Get an overview of ongoing AI safety and security standardization initiatives. We'll also discuss the impact of those new requirements to AI-based methodologies, architectures, and tools.

Discover the potential of NVIDIA's edge AI platform in transforming the future of smart edge and autonomous robotics. This session will provide an overview of the platform's capabilities across embedded edge, industrial edge, and enterprise edge environments. Learn about the best practices and software framework that enable AI-powered robots and intelligent vision systems from advanced sensor processing to real-time decision-making. Explore how these innovative technologies are creating generative physical AI systems and shaping industries across healthcare, retails, robotics, transportation, manufacturing and more.

The healthcare sector is evolving rapidly, with medical devices, treatment facilities, and operating rooms transitioning toward fully robotic systems. We'll address the vital role of physical AI — models designed to understand, interact, and navigate within this new paradigm. Delve into the comprehensive workflows across training, simulation, and runtime that underpin these technological advances. We'll explore how platforms like MONAI, Omniverse, Isaac Lab, and Holoscan are critical in developing next-generation medical devices. These tools create a robust ecosystem for innovation, empowering you to lead in the transformation of healthcare. Join us to enhance your skills and contribute to the future of medical technology, where intelligent robotic systems are the norm.

In this pivotal moment for industrial innovation, we explore how NVIDIA Robotics products — Omniverse, Isaac Sim, Metropolis, Fleet Management, and Discrete Event Simulation — are transforming warehouse and factory management from pre-commissioning to full-scale operations. We'll examine the overall value proposition of these cutting-edge technologies, focusing on their financial benefits, efficiency gains, and enhanced scalability. Additionally, we'll address the urgency for clients to adopt these technical enablers in response to evolving industry demands. Real-world examples will illustrate how these solutions meet the challenges of modern industrial environments, driving operational excellence and competitive advantage.

The world's largest industries are poised to embrace physical AI. The challenge comes with necessity to understand multimodally in real time. The entire sensor AI processing needs to be accelerated. Developers now can tackle edge AI challenges by seamlessly integrating sensor data, accelerated computing, real-time visualization/actuation, and AI inferencing — all while it abstracts away complexities for developers and reduces time-to-market. We'll feature cross-industry use cases including medical surgical robotics, brain-computer interface, semiconductor defect detection, and high-bandwidth sensor processing at scientific computing to demonstrate real-world physical AI applications. Cross-industry use cases will demonstrate real-world physical AI applications and sensor processing problems from model training and simulation to edge computing.