NVIDIA Physical AI Community

See. Do. Spark. 

Build AI that perceives the world, takes autonomous action, and pushes the limits of what's possible—all running on DGX Spark. 

NVIDIA is bringing the Spark Hack Series to Seattle.

This event brings together frontier builders, applied ML engineers, systems thinkers, and startup teams to explore what's possible when vision AI, agentic workflows, and open models meet high-performance local compute. Whether you're building systems that see, systems that act, or something no one's thought of yet—bring it. 

We'll provide every team with an Acer Veriton GN100—powered by the NVIDIA GB10 Grace Blackwell Superchip—to use throughout the weekend. 

Co-Host: Ascend backs pre-seed founders solving the biggest problems in enterprise AI and deep tech.

More details here: https://luma.com/spark-hack-seattle

Get warmed up for Actuate the best way possible—by building!

Join Nebius, NVIDIA, Antioch, and Toloka for a day of workshops, rapid experimentation, and building at the intersection of AI and the physical world.

We’ll begin with technical workshops designed to help you get familiar with the tools, infrastructure, and challenge space. Then, you’ll form a team—or work independently—and spend the rest of the day building a working prototype.

What to expect

• Robotics workshops
• A focused physical AI challenge
• Access to tools, infrastructure, and technical support
• Project demos and judging
• Prizes for the strongest submissions

Schedule (*subject to change)

• 8:00am: Doors open and check-in, grab breakfast
• 9:00am:  Welcome and challenge briefing
• 9:15am: Technical workshops by Hosts
• 10:30: Hacking begins
• 12:00pm: Lunch
• 1:00pm: Hacking continues
• 3:30pm: Submission deadline
• 4:30pm Demos and judging (top 6 live demos)
• 5:30pm: Winner announcement
• 6pm-9pm: happy hour (more details to come!)

The challenge

The Perception-Reasoning-Action Loop

Modern AI excels at generating text and pixels, but Physical AI requires closing the loop between the digital mind and the physical embodiment. Participants will build solutions that seamlessly integrate three core pillars:

1. Perception (Sense): Ingesting and interpreting multi-modal sensor data such as RGB-D video, LiDAR, IMU readings, tactile feedback, or spatial audio to build a real-time, dynamic understanding of the physical surroundings.
2. Reasoning (Think): Leveraging advanced foundation models, spatial computing, and decision-making frameworks to analyze constraints, anticipate outcomes, plan multi-step tasks, and adapt to unexpected real-world changes.
3. Action (Act): Translating high-level cognitive plans into precise physical outputs, including robotic manipulation, autonomous navigation, drone flight dynamics, or interactive IoT actuator responses.

Registration required, most details here: https://luma.com/nkknxvrz

Join NVIDIA at Actuate 26 on August 18–19, 2026, at Fort Mason in San Francisco for two days focused on the future of robotics and physical AI.

Connect with NVIDIA experts and the robotics community while exploring technologies and workflows spanning world foundation models, digital twins, synthetic data generation, and next-generation robotics development.

🛠️ Physical AI Sprint Hackathon | August 17
Kick off the week with the Physical AI Sprint Hackathon, featuring hands-on workshops, rapid experimentation, and the opportunity to build physical AI prototypes.

🔗 Learn more + register for the hackathon: here
📍 Explore NVIDIA at Actuate 26: here

Join OpenUSD Insider to explore how developers are using OpenUSD and NVIDIA Omniverse to build the next generation of physical AI. Each livestream features practical workflows, technical deep dives, and real-world examples spanning robotics, industrial digital twins, simulation, and autonomous systems.

Hear directly from developers and industry experts, discover the latest tools and resources, and learn how OpenUSD is helping teams create the physically accurate virtual environments needed to develop, train, and test intelligent machines.

Join our interactive Isaac Sim and Isaac Lab study group, led by NVIDIA Omniverse Ambassador Scott Dempsey. Explore robotics simulation, digital twins, and AI-driven control using NVIDIA’s advanced platforms and OpenUSD workflows. This hands-on series offers insights for all skill levels, fostering collaboration and practical learning for real-world robotics applications.

This weekly study group is held in the "Community Room" voice channel of the NVIDIA Omniverse Discord server:
https://discord.gg/wK52NSrhtw

For additional resources, please see our free self-guided Learning Path for Physical AI:
https://docs.nvidia.com/learning/physical-ai/index.html

The main Omniverse Discord server, which includes dedicated channels for the study group, as well as for Isaac Sim and Isaac Lab (be sure to complete the short form in the #apply-for-access channel to be approved as a member):
https://discord.com/invite/nvidiaomniverse

Schließe dich unserer gemeinsamen Study Group an, in der wir den OpenUSD Learning Path Schritt für Schritt durchgehen. So kannst du deine Fähigkeiten stärken, Erfahrungen mit anderen teilen und dich optimal auf die USD-Zertifizierung vorbereiten – alles in einer unterstützenden Lernumgebung.
Weitere Informationen: https://perfectproducts.github.io/openusd_studierstube/

The OpenUSD Study Group is a dedicated space for addressing challenges and issues encountered during OpenUSD learning. As a group, we share our experiences of how we solved some of the problems and challenges we faced while learning and building OpenUSD-based solutions.Additional Resources: https://ziacreatesideas.github.io/STUDY_USD-GROUP/index.html

Hosted by Vention

Most Physical AI demos show the easy parts. This session explains how to pick the hard ones.

Jimmy Li and Jean-François Tremblay open up GRIIP, the perception-to-motion pipeline behind Rapid Operator AI. From raw point clouds straight from the sensor, to how each model in the pipeline turns them into a reliable pick: segmentation, grasp selection, motion planning. The same models run every deployment, and the pipeline adapts to a new application in hours, not months. That architecture is why GRIIP holds up on the parts that break conventional vision: reflective, dark, transparent, tangled, or buried at the bottom of a deep bin.

The session covers how GRIIP segments cluttered scenes, selects grasps, and plans collision-free motion on real production parts. It also draws the line where classical vision stops working and foundation models such as NVIDIAs take over.

Attendees leave knowing exactly what runs inside the pipeline and why it holds up on a real factory floor. It comes down to one thing: hardware, AI pipeline, motion, and control engineered as one system, not stitched together from scattered vendors. It all runs on the Vention platform, the same platform behind 28,000+ machines. That is how automation stops breaking on real-world variability, and how it reaches the parts and the manufacturers it never could before.

Registration required: https://events.ringcentral.com/events/physical-ai-picking-the-impossible-parts-with-griiptm

Join OpenUSD Insider to explore how developers are using OpenUSD and NVIDIA Omniverse to build the next generation of physical AI. Each livestream features practical workflows, technical deep dives, and real-world examples spanning robotics, industrial digital twins, simulation, and autonomous systems.

Hear directly from developers and industry experts, discover the latest tools and resources, and learn how OpenUSD is helping teams create the physically accurate virtual environments needed to develop, train, and test intelligent machines.

Join our interactive Isaac Sim and Isaac Lab study group, led by NVIDIA Omniverse Ambassador Scott Dempsey. Explore robotics simulation, digital twins, and AI-driven control using NVIDIA’s advanced platforms and OpenUSD workflows. This hands-on series offers insights for all skill levels, fostering collaboration and practical learning for real-world robotics applications.

This weekly study group is held in the "Community Room" voice channel of the NVIDIA Omniverse Discord server:
https://discord.gg/wK52NSrhtw

For additional resources, please see our free self-guided Learning Path for Physical AI:
https://docs.nvidia.com/learning/physical-ai/index.html

The main Omniverse Discord server, which includes dedicated channels for the study group, as well as for Isaac Sim and Isaac Lab (be sure to complete the short form in the #apply-for-access channel to be approved as a member):
https://discord.com/invite/nvidiaomniverse

Schließe dich unserer gemeinsamen Study Group an, in der wir den OpenUSD Learning Path Schritt für Schritt durchgehen. So kannst du deine Fähigkeiten stärken, Erfahrungen mit anderen teilen und dich optimal auf die USD-Zertifizierung vorbereiten – alles in einer unterstützenden Lernumgebung.
Weitere Informationen: https://perfectproducts.github.io/openusd_studierstube/

The OpenUSD Study Group is a dedicated space for addressing challenges and issues encountered during OpenUSD learning. As a group, we share our experiences of how we solved some of the problems and challenges we faced while learning and building OpenUSD-based solutions.Additional Resources: https://ziacreatesideas.github.io/STUDY_USD-GROUP/index.html

The NVIDIA GTC AI conference is coming to Berlin, bringing together developers, researchers, and industry leaders to go deep on AI infrastructure. Sessions will span the entire five-layer AI stack, from energy, chips, and infrastructure to open models and physical AI applications empowering every industry and organization to build AI on their own terms.

See CEO Jensen Huang's live keynote at the iconic Tempodrom, then join us at STATION-Berlin for dynamic sessions, hands-on labs, and exclusive networking.