ovphysx

ovphysx is a self-contained Python library for ovstage/USD-based physics simulation with Warp-array output and downstream DLPack interoperability. It wraps NVIDIA PhysX and provides:

• ovstage attachment plus rigid-body / articulation simulation
• Same-device warp.array output on CPU and CUDA through session reads (PhysX.read() / read_sync())
• Downstream zero-copy framework interoperability through Warp's DLPack support
• Environment cloning for batched reinforcement-learning workloads

Quick start

pip install ovphysx physx-usd-schemas newton-usd-schemas

newton-usd-schemas is the Newton USD schema the snippet below registers next to the external PhysX schemas; scenes that author no newton:* attributes can leave it out and drop the newton_schema_root() entry.

from pathlib import Path

import ovphysx
from ovphysx.samples.python_samples.schema_support import physx_schema_paths
import ovstage
from ovphysx import PhysX

usd_path = (
    Path(ovphysx.__file__).resolve().parent
    / "samples"
    / "data"
    / "simple_physics_scene.usda"
)
if not usd_path.is_file():
    raise FileNotFoundError(f"ovphysx sample data is missing: {usd_path}")

# Register external PhysX and Newton schemas before the first population.
ovstage.population.register_usd_schemas(
    [*[str(p) for p in physx_schema_paths()], str(ovphysx.newton_schema_root())]
)
stage = ovstage.Stage("scene")
ovstage.population.open_usd(
    stage, str(usd_path), ordinal=1, domains=ovstage.PopulationDomain.ALL
)
# attach_ovstage() reads at a sealed ordinal.
stage.advance_write_floor(ordinal=1).wait()

physx = PhysX()
physx.wait_task(physx.attach_ovstage(stage, read_ordinal=1))
physx.step_sync(1.0 / 60.0)
physx.wait_task(physx.detach_ovstage())
physx.destroy()
stage.destroy()

For full documentation and tutorials, see ovphysx/docs/ inside the installed package, or the ovphysx repository.