Vehicle-in-the-Loop Testing Integration
Close the gap between HIL benches and on-road validation. We integrate real ECUs, dynamometers, and high-fidelity vehicle models into one testbed your team can actually run.
HIL benches don't replicate the vehicle. Road tests don't scale.
Modern automotive programs sit between two failure modes. HIL stops at the ECU boundary, so the vehicle dynamics, driver, and environment are simulated but never co-validated. Road test catches the rest, but it's expensive, slow, and impossible to repeat with the precision regulators and ASIL audits demand.
Vehicle-in-the-Loop fills the middle. A real ECU under test sees a real powertrain via a dynamometer, and the vehicle dynamics, sensors, and traffic environment run as a hard real-time simulation around it. You get bench reproducibility with road-level fidelity.
- Validate ECU behavior against high-fidelity vehicle dynamics
- Reproduce edge-case scenarios that road test can't safely run
- Run regression sweeps for ADAS and EV controls overnight
- Cut physical prototype iterations by 50%+ before SOP
- Generate ISO 26262 evidence with repeatable runs
Where VIL fits
HIL, VIL, and road test are not competing methods — they answer different questions at different fidelity, cost, and risk. The table below is how we frame the trade for validation leads.
| Dimension | HIL bench | VIL | Road test |
|---|---|---|---|
| What's physically real | The ECU and its I/O. Everything past the connector is a model. | ECU plus real powertrain hardware under dyno load; environment and traffic simulated in hard real time. | Everything — vehicle, driver, road, weather. |
| Vehicle dynamics fidelity | Model-limited; no physical load path, so plant-model error passes through untested. | High — real torque paths and actuator responses closed against a validated vehicle model. | Ground truth by definition. |
| Reproducibility | Excellent — fully deterministic and scriptable. | Excellent — same determinism, now with physical hardware in the loop. | Poor — weather, traffic, tire wear, and driver variance between runs. |
| Edge-case safety | Safe, but physical failure effects stay hypothetical. | Safe — crash-imminent maneuvers and injected faults run at full load with zero physical risk. | Limited — many required scenarios are too dangerous to stage. |
| Cost per scenario | Low once the bench exists. | Low to moderate after commissioning; marginal cost of a run is near zero. | High — prototype vehicles, drivers, instrumentation, proving-ground time. |
| Regression suitability | Strong for ECU software logic. | Strong for system-level behavior — CI-triggerable overnight sweeps against real hardware. | Impractical beyond spot checks. |
| Typical program stage | Early software integration and unit-level validation. | Mid-to-late integration through pre-SOP system validation. | Final validation, calibration sign-off, homologation. |
The honest read: a serious program needs all three. HIL catches software defects early, road test remains the final word, and VIL removes the expensive, slow gap between them — where most integration defects have historically hidden until a prototype vehicle found them.
What we integrate
VIL doesn't fail at any one piece — it fails at the seams between them. We focus on the integration.
ECU integration
CAN, CAN-FD, FlexRay, Automotive Ethernet, LIN. Restbus simulation, gateway routing, and timing-accurate signal injection for production ECUs. Intrepid neoVI interfaces and Vehicle Spy are our standard interface toolchain.
Dynamometer coupling
AC/DC dyno integration for powertrain load simulation, regenerative testing, and efficiency mapping. Closed-loop with real-time vehicle dynamics. Hub-coupled Rototest systems are our preferred approach where tire variability must be eliminated.
Scenario authoring
OpenSCENARIO and OpenDRIVE-driven test cases. Sensor simulation for camera, radar, LiDAR with traffic, weather, and driver model variation.
Data pipeline
High-rate signal capture, MDF4 logging, automated post-processing, and CI-friendly reports. Your test sweeps end in a comparable artifact, not a folder of CSVs.
Works with your existing toolchain
Prysm8 is an official partner of Toyotech Rototest for hub-coupled dynamometer systems, works closely with Intrepid Control Systems on vehicle networking, and remains vendor-neutral across the rest of the stack. We integrate what you already have. Common integrations include:
How we work
Three phases, scoped before we start, with deliverables you can hand to your validation lead.
Discover
We map the test gap: which ECUs, which scenarios, which signals, which compliance evidence you need. Output is a concrete VIL architecture document and an integration plan.
Integrate
We wire your ECU under test into the dyno and real-time vehicle model, validate signal timing, and build the scenario library. We work with your existing simulation hosts whenever possible.
Validate
We hand over a running testbed, an automated regression suite, and documentation. Your team owns it; we stay available for the next program.
Recent engagements
Representative programs. Client identities and program details are generalized under NDA.
HIL testbed with dynamometer integration for a hybrid program
Built a powertrain HIL testbed around dSPACE SCALEXIO with real-time engine and e-motor plant models coupled to dynamometer load control. Full restbus simulation across CAN, FlexRay, and Automotive Ethernet, with automated fault injection at the signal and bus level.
The bench replaced a large share of prototype-vehicle test hours: torque-path faults, limp-home transitions, and hybrid mode arbitration were validated under repeatable load before a vehicle existed.
Multi-domain validation rig for an ADAS platform
Integrated a multi-domain rig with camera, radar, and LiDAR simulation driven by IPG CarMaker. Traffic scenarios authored in OpenDRIVE and OpenSCENARIO, executed as automated regression sweeps against the production sensor-fusion and planning ECUs.
Deliverables included the scenario library, CI integration, and traceable evidence supporting the client's ISO 26262 ASIL-D safety case.
Common questions
How long does a VIL integration take?
Phased. The architecture and integration plan lands in weeks — that document alone tells you what the full build costs and where the risks are. Full integration is scoped per program: an extension to an existing bench is a different timeline than a ground-up rig with dyno commissioning. We quote the phases separately so you can stop after any of them.
Do you work on-site or remote?
Both. Commissioning, dyno coupling, and hardware bring-up happen on-site at your facility. Model development, scenario authoring, and test automation are done remotely. We're based in Greater Detroit, so on-site work across the Midwest automotive corridor is routine.
Who owns the testbed and IP after handover?
You do — hardware configuration, models, scenario libraries, automation scripts, and documentation are all deliverables. There is no proprietary runtime, no license tied to us, and no lock-in. The handover criterion is that your team can run and extend the bench without us.
Can you extend an existing HIL bench instead of building new?
Yes, and most engagements start exactly there. Adding dyno coupling, higher-fidelity vehicle models, or sensor simulation to a bench you already own is usually the fastest path to VIL — and it preserves the test assets and operator knowledge your team has already built.
How do you handle NDAs and export-controlled data?
Standard practice. We execute NDAs before any technical exchange, and we discuss export-control classification (EAR/ITAR where applicable) up front so data handling, access, and storage are settled before the first file moves.