Rehearse the facility before you build it¶

Deployment rehearsal Doc torus-isaac-sim Class Cyber-physical twin Real NVIDIA Isaac Sim and OpenUSD output

A TORUS deployment is rehearsed as a cyber-physical twin. Terrain is reconstructed from mapping data, the full system is placed virtually, and unauthorized-entry scenarios are run against it before anyone digs a trench.

Twin

OpenUSD campus

Terrain, halls, roads, perimeter, water

Planner

Offline coverage

Set-cover / art-gallery approximation

Runtime

Isaac Sim

Physics and sensor simulation

Output

Commissioning plan

Node positions, thresholds, scenario evidence

TORUS repository torus-isaac-sim repository Imagery library User and developer guide Parameter reference Build, deploy, and use

Everything on this page is simulation output

The images on this page are captures from the OpenUSD campus generator and NVIDIA Isaac Sim scenes in torus-isaac-sim. They are not photographs, and they are labelled separately from the concept collateral shown in the imagery library.

NVIDIA credit

NVIDIA Isaac Sim, OpenUSD, Omniverse, PhysX, Jetson, and Orin are credited as NVIDIA technologies used or referenced by the TORUS deployment rehearsal.

01 · Why rehearse the facility¶

Optimal sensor placement for complete coverage is a set-cover and art-gallery problem, which is NP-hard. TORUS does not pretend otherwise. The coverage problem is approximated off-line in simulation, and only the zero-miss processing guarantee is made on-line. AI/HPC campuses are expensive to survey twice, so the plan is validated in the twin and then promoted into the live CCISRT picture.

In data center campus deployments the twin helps twice. The local noise floor is simulated first, so seismic thresholds and acoustic classifiers arrive on site pre-tuned against cooling plant, generator, and traffic background instead of by trial and error.

Constraint	Simulation use
Coverage is hard	Approximate the set-cover / art-gallery problem off-line before committing hardware.
Survey cost	Validate ring geometry, visual coverage, and access routes before the field visit.
Local noise	Pre-tune seismic thresholds and acoustic classifiers against the simulated site background.
Operations handoff	Promote the commissioned plan into the live CCISRT picture.

02 · Terrain and twin construction¶

The generator builds a layered AI data center campus in OpenUSD: halls, infrastructure, habitat areas, roads, perimeter, and retention water. It is portable and dependency-free for planning, and loads into full Isaac Sim for physics and sensor simulation.

Aerial OpenUSD campus view from the southeast — Simulation output: layered data center campus with halls, infrastructure, habitat areas, roads, perimeter, and retention water.

Aerial OpenUSD campus view from the northwest — Simulation output: opposite approach, used for sight-line and access-route checks.

Isaac Sim main gate approach at the campus perimeter — Simulation output: main-gate approach for placing ground sensing, vehicle classification, and visual confirmation coverage.

03 · Sensor placement and RF planning¶

The planner places the ground ring, the gateways, and the visual masts, then evaluates coverage against the object classes the site cares about: human, vehicle, drone, and animal. All three UAGVIS domains are rehearsed in the same scene, so air, ground and visual coverage are planned against one another rather than in isolation.

Isaac Sim overview of a TORUS ground ring, visual masts, wireless telemetry, and a drone scenario — Simulation output: multi-domain campus rehearsal combining the ground ring, SN-TIR masts, RF telemetry, and a low-altitude UAS track.

Isaac Sim view of the TORUS instrumented campus perimeter — Simulation output: instrumented campus perimeter with the TORUS-SN ground ring in place.

OpenUSD campus showing TORUS RF fields and telemetry paths — Simulation output: scenario-controlled RF-field and telemetry planning view. It visualizes link assumptions; it is not a full-wave electromagnetic solution.

04 · Procedural asset set¶

Nodes, gateways, and masts are procedural assets in the scene, so a scenario can move them, disable them, or fail their links and observe what the operator would actually have seen. The assets track the current industrial design, so the rehearsal shows the hardware that will actually be emplaced.

Simulation asset: current TORUS-SN design language, low-profile emplaced enclosure, three RF whips, status indication, and service lead.

Procedural pole-mounted TORUS-MEG gateway with solar panel — Simulation asset: current TORUS-MEG design language, pole mounting, solar power, local display and controls, dual radio antennas, and battery module.

Procedural TORUS-SN-TIR thermal visual mast — Simulation asset: current SN-TIR design language, six-metre mast with thermal/day optics, solar power, radio, and electronics enclosure.

Isaac Sim view of the TORUS-MEG gateway service and facility handoff area — Simulation output: gateway service and handoff view for maintainability, health checks, and facility backhaul planning.

05 · Scenario and sensor simulation¶

Campus-perimeter and utility-yard scenarios exercise the four object classes. The dependency-free planner produces ring placement, a seismic detection model, and SVG output without Isaac Sim installed; the full twin adds PhysX-based detection and visual-mast sensors.

Labelled TORUS wireless sensor deployment zone — Simulation output: TORUS-SN nodes, TORUS-MEG aggregation, geospatial coverage, and wireless telemetry in one labelled zone.

Illustrative TORUS mast thermal camera feed — Simulation output: illustrative ironbow thermal display from MAST-01 with the selected actor in view. Not calibrated radiometric output.

Isaac Sim view of a perimeter intruder route and correlated TORUS detections — Simulation output: intrusion-correlation view with the active route and multi-node detection evidence around a protected geofence.

Isaac Sim rehearsal of a classified delivery truck approaching the data center gate — Simulation output: authorized commercial-vehicle approach with route context for gate classification and policy testing.

Isaac Sim rehearsal of a hostile low-altitude drone crossing the TORUS campus — Simulation output: low-altitude UAS rehearsal for correlating air-domain cues with ground and visual sensing.

Output	Use
Node positions and ring geometry	Promoted directly into the live CCISRT picture as the commissioned plan.
Per-node gain and threshold settings	Pre-tuned against the simulated local noise floor rather than tuned by trial on site.
Detection and false-alarm rates per class	Acceptance evidence for the coverage design before hardware is committed.
Labelled scenario data	Training and validation data for the edge classifiers and the CCISRT correlator.
First-contact timing	Sample run: 8 nodes, 250 m ring, first contact at t+49 s on node SN-06, 4 nodes triggered along the track.

06 · Site-fidelity checks¶

The twin keeps the protected asset, retention water, green buffers, and approach routes in the same scene so sight lines, access routes, material-loss assumptions, seismic coupling, and classification context can be reviewed together.

Isaac Sim view of data hall blocks on the campus — Simulation output: data hall block, the asset the perimeter ring exists to protect.

Isaac Sim view across the west retention lake — Simulation output: west retention-water and green-buffer environment for testing sight lines, access routes, and material-loss assumptions.

Isaac Sim view across the east retention lake — Simulation output: east retention water. Standing water changes both seismic coupling and approach routes.