NVIDIA GTC 2026
ADI AT NVIDIA GTC ADI AT NVIDIA GTC
Smarter Robots. Sharper Insights. Physical Intelligence Unlocked.
ADI brings physical intelligence to humanoid robots. We unite sensors, motor control, real time connectivity, AI, and high fidelity simulation in a software defined humanoid hand. The result: faster design cycles, reliable sim to real transfer, and a clear path from prototype to scalable, manufacturable robotic hands.
Hear From Our Experts
Paul Golding
Vice President, Edge AI
The Edge of the Edge: Redefining GPU-Enabled AI Sensor Processing [S81525]
Analog Devices, Inc. (ADI), Physical AI and edge supercomputing are both concerned with the timely delivery and processing of physical measurements on GPU to build the future of autonomous decision-making across industries and domains—from scientific instruments to industrial robotics. During this talk, we will hear from leading sensor manufacturers who are building real-time AI solutions with NVIDIA and the Holoscan platform; they'll share their best practices and their vision, and highlight transformational work toward a future of AI-powered instruments.
Philip Sharos
Principal Engineer, Edge AI
Reducing the Sim2Real Gap for Industrial Robotics
Robotic systems increasingly demonstrate strong capabilities, yet still struggle with real world dexterity due to limitations in sensing, actuation, and high quality training data. This session, presented by Synopsys and ADI, discusses the critical need for new sensing modalities, and high fidelity simulation techniques that accelerate learning and significantly narrow the Sim to Real gap for industrial manipulation. The presentation will also touch on ADI’s work on novel dexterity benchmarks and Synopsys contributions to deformable objects and connector assets.
Tao Yu
Director, Dexterous AI, Emerging Business Unit
Bringing Physical Intelligence into Scalable Robotics Simulation: Real2Sim Validation for Industrial Robotics
This talk presents a joint effort between Analog Devices and Lightwheel to improve simulation fidelity for high precision robotic tasks. The work focuses on a real2sim validation pipeline that compares measured forces, contact behavior, and deformable interactions against simulated results. By pairing advanced sensing with a calibrated physics engine, the collaboration provides a clear way to evaluate where simulation aligns with physical measurements and where it does not. The goal is to support more dependable training and evaluation for manipulation systems that depend on subtle physical feedback.
Preview Our Demos
Tactile Intelligence: ADI’s Dexterous Humanoid Hand
ADI’s software defined humanoid hand demonstration platform combines our next generation tactile sensor with a precision signal chain plus motor control to deliver unmatched contact accuracy and control. New sensing modalities, enhanced dexterity, and AI-ready simulation assets accelerate prototyping, boost performance, and provide a clear, scalable path from early concepts to manufacturable robotic systems.
Learn More:
Bringing Human Touch to Robots: The Future of Tactile Sensing
800 V to Core Power for AI
Analog Devices is driving the transition to 800 V rack-level power distribution for next-generation AI data centers by focusing on high-voltage protection and conversion, power delivery, and advanced control technologies. As high-performance GPUs drive unprecedented power and current demands, the industry is shifting from legacy 48 V architectures to 800 V systems. ADI’s advanced high-voltage hot-swap technology enables a safe and efficient connection from 800 V down to the GPU core, with precise inrush control, comprehensive telemetry, and robust protection. These high-density “800 V-to-core” power solutions—spanning 800 V distribution and protection, high-voltage DC/DC conversion, intermediate bus converters (IBC), and smart multiphase load regulation—deliver exceptional power density, support massive currents, and ensure the reliability required for modern AI data centers.
Learn More
AI-Driven Humanoid Hand Cable Manipulation
Building a robotic signal chain for humanoid hands is the starting point. Achieving reliable dexterity requires selecting the right sensing and actuation components, integrating them into a coherent signal chain, and developing control policies that can perform under real world constraints. Advanced manipulation depends on AI driven policies shaped by both manual demonstrations and synthetic data. ADI reduces this complexity with physics aligned sensing, mixed signal compute and a robotics policy stack that accelerates sim to real development. These capabilities are demonstrated on our software defined humanoid hand platform.
Learn More
Sim-to-Real Robotic Dexterity Benchmark at the Synopsys booth
ADI, together with our partners at Synopsys, is showcasing a simulation-driven recreation of our next-generation robotic dexterity benchmarking tests. The demo integrates ADI’s next generation Tactile Sensors with Time-of-Flight sensing and Ansys Mechanical multiphysics simulation to enable advanced intelligent cable manipulation. It combines a bi-manual robotic arm setup from Flexiv with force, vision, and contact sensing, all connected within a high-fidelity digital twin to deliver predictive, optimized, and robust performance in dynamic industrial environments.
Real Time Dexterity Benchmark Simulation
ADI and its partners are working on a new benchmark to help the industry objectively assess robotic dexterity solutions and accelerate innovation at scale. Complemented by datasets and AI simulation assets, this effort was showcased through the first test bench and its interactive digital twin at our booth.
Learn More