Skip to navigation

Getting started

View as Markdown

Run a sample application without a robot or simulator. You will check your host, fetch the source code and run so101_sinus_motion mock to generate joint commands from synthetic states. This introduces the same build-and-run workflow used by the robot applications.

Check your host

Ensure you have a host that meets the platform requirements. AGX Thor with JetPack 7.2.1 is the formally supported deployment platform; x86_64 with NVIDIA GPU and AGX Orin on JetPack 7 are expected to support this mock workflow also.

Check the tools needed to run the sample:

git --version
docker info --format '{{json .Runtimes}}'
uv --version

Git and uv should print their versions. Docker must work without sudo and list nvidia among its runtimes. If a command fails or nvidia is missing, complete the Build and run prerequisites, then return here.

Fetch the reference source code

git clone --branch v0.11.0 https://github.com/nvidia-isaac/isaac-os.git
cd isaac-os

Run a sample app

./isaac_os run so101_sinus_motion mock -- --message-count=100

The first run downloads the container image and dependencies, then compiles the application. Later runs reuse the image, dependencies and build cache, so they start faster. This sample needs no policy model or TensorRT engine.

What to expect

The application prints changing six-joint targets and stops after 100 completed commands. The synthetic state source, sine generator and motion planner produce those targets; no robot moves.

With the default 100 Hz configuration, the log includes these messages. Prefixes, trace IDs, latencies and target values are omitted here:

Waiting up to 13.0 s for 100 SO101 sinus commands.
SO101 sinus mock command 10 received: ...
SO101 sinus mock command 20 received: ...
...
SO101 sinus mock command 100 received: ...

Next steps

Try another mode or application

List applications and the modes of a selected application:

./isaac_os list
./isaac_os modes so101_sinus_motion

Browse Controllers and Samples under Reference for application details, or try SO-101 motion in Isaac Sim.

Choose what to learn next