Toy train

Articulations

Push a toy train along a track on its free-spinning wheels.

Play0.00 s

¼×½×1×2×

Loop

The steel preset (the default), recordedon an NVIDIA RTX PRO 6000 Blackwell.

Run it

cd examples
# the default preset (steel), then the others
uv run shelf play toy_train
# change solver settings live
uv run shelf tune toy_train

Source: examples/shelf/07_articulations/toy_train/ · ported from steel-webgpu assets/scenes/toy_train (hero_montage export) @ ad03e49

Presets

Preset	Solver	Status	Checked
steeldefault	steel	works	2026-11-10, newton f6fc82a8b, NVIDIA RTX PRO 6000 Blackwell Server Edition

Solvers with no row are untested on this example. The full settings are inpresets.toml.

Code

example.py

examples/shelf/07_articulations/toy_train/example.py

# SPDX-FileCopyrightText: Copyright (c) 2026 Skild AI and the Steel contributors
# SPDX-License-Identifier: Apache-2.0
# Direct source: steel-webgpu; see presets.toml and THIRD_PARTY.md.
# Modified for Steel Shelf: preset-driven execution and shared runner integration.

###########################################################################
# Example Toy train
#
# The arm pushes a jointed toy train along a lane; its wheels spin freely as it rolls.
#
# Rebuilt from steel-webgpu's exported scene (assets/scenes/toy_train: scene.json
# and scene.bin, copied next to this file); see common/exported_scene.py. The
# visual meshes draw when a steel-webgpu checkout sits next to the shelf.
#
# Command: python -m shelf run toy_train
#
###########################################################################

import newton.examples
from common.exported_scene import ExportedSceneExample


class Example(ExportedSceneExample):
    file = __file__


if __name__ == "__main__":
    parser = Example.create_parser()
    viewer, args = newton.examples.init(parser)
    newton.examples.run(Example(viewer, args), args)

example.py on GitHub

presets.toml

examples/shelf/07_articulations/toy_train/presets.toml

example = "toy_train"
source = "steel-webgpu assets/scenes/toy_train (hero_montage export) @ ad03e49"
default = "steel"

[check]
# Arguments `python -m shelf check` passes in test mode.
args = []

[gallery]
short = "Push a toy train along a track on its free-spinning wheels."
title = "Toy train"
frame = 420

[xpbd]
status = "untested"

[vbd]
status = "untested"

[mujoco]
status = "untested"

[featherstone]
status = "untested"

[semi_implicit]
status = "untested"

[feather_pgs]
status = "untested"

[steel]
status = "works"
# 8 substeps x 2 iterations, immediate response. The substep count decides this scene: the train coasts on 24-sided
# wheel hulls, which lose more energy per facet at large steps. With 4 or fewer substeps (the browser's 4 x 16
# included) it stops 0.50-0.58 m down the lane, right at the tool, and when the tool lifts away at ~8 s it can
# snag the train and lift it 4-8 cm (seen in 5 of ~45 such runs, all with the train short of 0.52 m; 6 x 4 too).
# At 8 substeps it coasts to 0.62-0.68 m (converged 16 x 32 / 32 x 48: 0.63-0.69), clear of the tool in all 9
# runs, wheels rolling without slip (within 1%), peak penetration 0.1 mm. Iterations barely cost here (8 x 4:
# +0.14 ms). Velocity limits off: every exported limit is 1e6. Rows: at most 50 used.
sim = { fps = 60, substeps = 8 }
solver = { solve_mode = "matrix_free", iterations = 2, articulated_contact_response = "immediate", enable_joint_limits = true, joint_limit_activation_gap = 0.1, enable_joint_velocity_limits = false, articulation_constraint_max = 128, free_body_constraint_max = 128 }
checked = "2026-11-10, newton f6fc82a8b, NVIDIA RTX PRO 6000 Blackwell Server Edition"

[kamino]
status = "untested"

[implicit_mpm]
status = "untested"

[style3d]
status = "untested"

presets.toml on GitHub

← ToasterAll examplesFranka cable ik pick place →