Physical AI: The Messy Reality of Field Autonomy

About This Event

​How do we move beyond standard simulation to create high-fidelity digital twins that truly mirror the real world?

​Simulation is critical for modern robotics, but the “sim-to-real” gap remains a significant bottleneck, especially in complex environments involving fluids, deformable objects, and chemical processes.

​Join the Canadian Physical AI Institute (CPAI) for an exclusive session with Kourosh Darvish, Staff Scientist at the University of Toronto, as we explore Matterix—a new multi-scale, GPU-accelerated simulation framework designed to create high-fidelity digital twins.

​While originally developed for chemistry laboratory automation and self-driving labs, the concepts covered are broadly applicable to robotics domains requiring complex manipulation and perception.

​🚀 What You Will Learn

​This session is designed to provide both a “behind-the-scenes” look at real research projects and a practical walkthrough of key concepts.

• ​Introduction to Matterix: Discover a framework built on NVIDIA Isaac Sim and Isaac Lab that integrates realistic physics and photorealistic rendering with a modular, GPU-accelerated semantics engine.

• ​Multi-Scale Modeling: Learn how to simulate beyond simple rigid bodies, including powder and liquid dynamics, device functionalities, heat transfer, and chemical reaction kinetics.

• ​Hybrid Intelligence: Explore how to model both logical states and continuous behaviors to represent workflows across multiple levels of abstraction.

• ​Digital Twin Value: Gain a clear view of where digital twins add value, specifically in policy training, workflow development, system co-design, and safety analysis

• ​Real-World Tradeoffs: An honest discussion on when simulation works well, when it breaks down, and how to navigate those tradeoffs in real projects.

​📅 Agenda (ET)

​6:00 PM | Part 1: Introduction to Matterix

• ​Deep dive into creating high-fidelity digital twins using open-source asset libraries and standardized interfaces.

• ​How to enable flexible workflow design via hierarchical planning and modular skill libraries.

• ​Examples of modeling complex material interactions, from robot manipulation to perception.

​6:45 PM | Part 2: Simulation for Robot Learning

• ​How Matterix and Isaac Sim support data generation for robot learning workflows.

• ​Addressing cases where standard simulation abstractions fall short in lab environments.

​7:30 PM | Open Discussion & QA

• ​A hybrid discussion involving academia and industry on the “real-to-sim” gap.

• ​Key Discussion Points:

  • ​Creation of high-fidelity assets using generative models and 3D computer vision.

  • ​Augmenting simulation with synthetic data and world models to improve realism and transferability.

  • ​How industry practitioners can leverage digital twins for design, testing, and maintenance.

​👨‍💻 Who Should Attend?

​This event is intentionally designed to be useful for both beginners and experienced users:

• ​Beginners: Gain exposure to modern robotics simulation pipelines and current research directions.

• ​Industry Practitioners: Those looking to add flexibility, generalization, and faster iteration to real-world robotic systems.

• ​Researchers & Students: Anyone working on robot learning, digital twins, embodied AI, or self-driving laboratories.

​Meet Your Speakers

​Alec Krawciw

​Alec is a fourth-year PhD candidate at the University of Toronto Autonomous Space Robotics Lab and a Vanier Scholar. Holding an undergraduate degree in Mechanical Engineering from the University of Victoria, his work focuses on developing and testing autonomous vehicles in unstructured environments. His field experience ranges from water and snow to lunar analogue environments, and his current PhD research centers on coordinated driving algorithms for Canada’s upcoming Lunar Utility Vehicle.

​Behnam Moradi

​Behnam is a Senior Software Engineer specializing in the architectural evolution of Robotic Autonomy and Embodied AI. Holding a Master’s in Control Systems Engineering, his career marks a progression from deterministic control systems to the dynamic, unstructured challenges of autonomous navigation. His current work focuses on moving beyond rigid, hard-coded state machines and Behavior Trees toward priority-driven, mission-oriented agents capable of real-time decision-making in complex environments.

​Hosted By: Diana Gomez Galeano

​Diana studied Mechanical Engineering at McGill University and previously served as Director of McGill Robotics. She has moderated and hosted numerous technical events, bringing together researchers, engineers, and students to explore the frontiers of emerging technologies, robotics, and Physical AI.