Getting Started
Five minutes from clone to a controlled reach.
Prerequisites
- Python 3.12 or higher.
- The
uvpackage manager (recommended; plainpip install .also works).
Install
1. Pose a robot (kinematics)
Open the kinematics inspector on a bundled 4-DOF arm — move the joint sliders (forward kinematics) or click/drag the tip in the canvas (inverse kinematics):
2. Simulate its dynamics
Launch the real-time dynamics simulator. The window opens paused: press
Space (or the play button) to start, then press and drag in the canvas to pull
the tip with a spring force, drawn as a red arrow:
3. Run a controlled reach
Run a complete scenario — robot, start pose, task, and controller from one TOML file. The controller drives the arm to the purple target; drag the tip to disturb it and watch it recover:
Every simulator window shares the same transport bar (Space play/pause, →
step while paused, R reset, Q quit) and records the run — press Export…
to save a *.sklog.npz log, then replay it:
Where next
- Configure a robot of your own.
- Run controlled scenarios — the scenario file model, tasks, and controllers.
- Record, replay, and re-simulate — what the logs capture and what "reproducible" means.
- Python API quick start — use the library from code.