Is there an Extension planned or being developed for FMI/FMU Support in Omniverse?
There is no official FMI/FMU (Functional Mock-up Interface) extension currently available or planned for Omniverse that I’m aware of.
The search results show various physics simulation capabilities in Omniverse (PhysX vehicle dynamics, surface velocity APIs for conveyor belts, etc.), but no FMI/FMU support.
Possible Alternatives:
- Custom Extension Development - You could develop a custom Kit extension that:
- Imports FMU models using Python FMI libraries (like FMPy or PyFMI)
- Integrates the co-simulation into Omniverse’s timeline
- Maps FMU inputs/outputs to USD attributes
- PhysX Integration - For mechanical systems, Omniverse’s PhysX simulation might cover some FMU use cases directly through:
- Articulations and rigid body dynamics
- Custom force/torque application
- Vehicle dynamics APIs
- External Co-simulation - Run the FMU simulation externally and sync with Omniverse via:
- LiveSession for real-time updates
- USD attributes updated through Python scripting
- Custom message passing integration
What about this announcement : Using OpenUSD for Modular and Scalable Robotic Simulation and Development | NVIDIA Technical Blog ? “USD + FMI proposal: Co-standardization and conceptual data mapping between OpenUSD and FMI for co-simulation (such as multiple solvers simulating different components of the same robot or its surrounding environment).”
Has there been progress ?
I think there is clearly a need.
Besides nvidia predveloment (see e.g. Build Physics-Based Digital Twins for Co-Simulation (Presented by SoftServe) S71963 | GTC 2025 | NVIDIA On-Demand ) , third parties are creating their own solutions, see e.g. https://www.youtube.com/watch?v=Ned6kKgTUcc
I have a CAD model for Hydraulic actuators in USD composer and a Hydraulic circuit model in Modelica using OpenModelicaEditor. I am trying to use the FMU exported from Modelica tool to provide actuation to the actuator models. The FMU I am able to generate from OMEdit is unfortunately Model Exchange. Currently, I am using PyFMI to run the FMU, though havent found a way to run it in real-time.
I can ask about this and get back to you.
I have a reply from the NVIDIA Engineers. Currently, what we have is experimental, which is what we presented as a ‘proposal’ last year in GTC. Work is continuing on this.
This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.