Digital twins and simulation

DATA INFRASTRUCTURE FOR PHYSICAL AI.

Scalable digital twins and simulation-ready environments for the next generation of intelligent systems.

See how it works

One connected workflow

From a real place to usable autonomy data.

SimForge keeps maps, scenarios, simulation runs, and generated artifacts connected so teams can reproduce results instead of rebuilding context.

  1. 01

    Prepare a map

    Import or select a digital twin with lane topology, signals, and runtime metadata.

  2. 02

    Author scenarios

    Place actors, define behavior, and organize repeatable cases into datasets.

  3. 03

    Run simulation

    Dispatch CARLA jobs and capture synchronized RGB, depth, segmentation, and LiDAR.

  4. 04

    Export data

    Review outputs, package datasets, and move validated evidence into training or evaluation.

Simulation run · Sensor reviewCompleted
SimForge synchronized RGB, depth, segmentation, detection, LiDAR, and top-down sensor outputs

Trusted by research and mobility teams

UC Berkeley
California PATH
Path2V2X
NVIDIA Inception Program
// 002 — SimCloud

The simulation
platform.

Compose scenarios in the browser, run them on CARLA workers, and capture multi-sensor output for every run.

RGB
DEPTH
SEG
car · 0.93BOXES
Sensor Output · Synchronized Frames

NVIDIA Cosmos Transfer

One render,
every condition.

Push CARLA renders through NVIDIA Cosmos. Same actors, same trajectory — photorealistic variants for every weather, time of day, and lighting preset.

01 Generated outputs
Loading generated outputs…

// 004 — Cosmos-Reason2 VLM

Scenes the model
can explain.

Trailing-camera footage flows through Cosmos-Reason2. Ask a question about any moment in the scene and get structured commentary on what the model sees, beat by beat.

01 Live commentaryCosmos-Reason2
What is happening in this scene?
Loading footage
Trailing camera01
You
A yellow Tesla drives into a parking lot and takes various turns in the lot, eventually exiting out the way it initially turned in.
Cosmos-Reason2

Use cases

Built for teams shipping physical AI.

Autonomous driving wireframe visualization
01AV

Autonomous driving

Validate stacks against real street layouts as twin-backed, repeatable CARLA scenarios.

Physical AI labs wireframe visualization
02Robotics

Physical AI labs

Reproduce edge cases as deterministic runs your fleet and policies can re-run on demand.

Smart-city ops wireframe visualization
03Cities

Smart-city ops

Test traffic, curbside, and signal changes in-engine before any field trial.

Gaming, movies, entertainment wireframe visualization
04Media

Gaming, movies, entertainment

Turn real-world digital twins into cinematic previs, virtual sets, and playable worlds.

Custom maps & scenarios

Custom digital twins, authored scenarios, ready to run.

Send us a real-world location and the scenarios you want to run. We build the high-fidelity digital twin, author the scenarios, and hand everything back ready to run on the simulation platform.

See the workflow

SimForge · Building #159, 1301 South 46th St, Richmond, CA 94804 USA

SimForge — Data Infrastructure for Physical AI