SFR in Action

How Classification Works

Four questions that determine which tier a simulator belongs to

No methodology jargon. No standards language. This page explains classification the way you would explain it to someone who has never encountered the framework before, using plain descriptions of what is actually being asked at each step.

One Thing Is Being Determined


Every classification question, every piece of evidence, every criterion in the formal methodology is ultimately working toward a single determination:

The Central Question
Does the motion the participant feels correspond directly to what the vehicle is doing in the physics simulation — or is it something else?
If yes: the participant's inner ear receives information that matches the physics state. If no: the inner ear receives a motion signal that does not correspond to vehicle reality. That difference is the entire classification system.

All four questions below refine this core question. The first asks about the source of the motion. The second asks about the delivery path. The third asks about what actually arrives. The fourth asks how we know. Together they narrow down to one of three classification results.

1
Where does the motion come from?

A simulator has a physics engine — software that calculates what the vehicle is doing at every moment: how much it is turning, braking, accelerating, rolling, pitching. Question 1 asks: does the platform's motion come directly from those calculations, or does it come from somewhere else?

If yes
Motion comes directly from the physics engine output
The actuator receives a command that was generated by the physics calculation at the moment it occurred. Proceed to Question 2.
If no
Motion comes from a motion-effect layer or pre-programmed profiles
The actuator receives a command from a motion cueing algorithm, a washout-processed signal, or pre-scripted motion profiles. Classification is Surface-Level. No further questions needed for classification.
If unsure
Architecture not documented or not provided
Without architecture documentation, Criterion A cannot be assessed. Result: Insufficient Data. Classification cannot proceed.
Signal Flow: Physics-Derived vs. Effect-Layer

Physics-derived (passes Question 1):

Vehicle physics model
Actuator command
Motion reaches participant

Effect-layer (fails Question 1):

Vehicle physics model
Motion cueing algorithm / washout filter
Transformed signal
Motion reaches participant
The washout filter does not reduce the motion — it changes what the motion represents. The participant receives a motion that is an approximation of the physics state, not the physics state itself.
2
How is the motion delivered?

Even if motion originates from the physics engine, it has to travel to the participant through the mechanical structure of the platform. Question 2 asks whether that delivery preserves the physics-derived signal correctly — or whether the mechanical architecture distorts it.

Two things matter here: whether each axis of motion (forward/back, left/right, up/down, and three rotational directions) operates independently, and whether the rotational calculations are referenced to the vehicle's center of mass.

If independent
Each axis has its own actuator, calculated from the vehicle center of mass
The physics signal for each axis reaches the participant without cross-contamination from other axes. Proceed to Question 3.
If coupled
Axes share a mechanical resolution (hexapod, shared kinematics)
Even if motion originated from the physics engine, the mechanical coupling means no axis reaches the participant cleanly. Classification is Surface-Level.
If unsure
Actuator architecture not documented
Without actuator specification, structural independence cannot be confirmed. Result: Insufficient Data on Criterion B.

The hexapod is a common platform configuration. It achieves six degrees of freedom through six actuator legs — but all six DOF are resolved simultaneously through the kinematics of those six legs. There is no independent axis. A command to produce yaw rotation requires coordinated changes across all six legs. This is mechanically inherent to the hexapod geometry and cannot be designed around without changing the fundamental configuration.

3
What reaches the participant?

Questions 1 and 2 examine the architecture. Question 3 examines the result. Even if a system passes Questions 1 and 2 on paper, the actual motion characteristics — timing, magnitude, onset pattern — need to be consistent with what the inner ear would expect to detect for the vehicle events being simulated.

This is not about comfort or preference. The inner ear has specific detection characteristics: it responds to acceleration onset above a threshold, it detects direction and magnitude, and it relates timing to visual input to assess whether what it feels matches what it sees. If the motion timing or magnitude are too far outside the expected range for the depicted vehicle events, the brain receives a mismatched signal even if the architecture is correct.

If consistent
Motion timing and magnitude match the depicted vehicle event parameters
The inner ear receives a valid, physics-corresponding signal. All three criteria are satisfied. Classification is In-the-Loop.
If inconsistent
Motion is attenuated, delayed, or misrepresents the vehicle event
The participant receives a motion signal that does not correctly represent the vehicle physics state. Criterion C fails. Classification depends on the overall criterion summary.
If not measured
No telemetry or timing data available
Without measured motion data, Criterion C cannot be assessed from direct observation. Result: Insufficient Data on Criterion C.
4
What evidence supports the answers?

A classification is only as reliable as the evidence behind it. Question 4 asks: how was each of the first three questions actually answered? Was it from directly measured data, from verified technical documentation, from manufacturer-provided specifications, or from claims in a product brochure?

These are not equally reliable. The framework assigns four evidence tiers to distinguish them. The classification result does not change based on the evidence tier — a Pass is a Pass regardless of tier — but the confidence in the result does.

1
Measured Telemetry
Directly recorded actuator and physics data. Highest quality. Answers the question from observation.
2
Verified Documentation
Engineering records reviewed and confirmed. High quality. Describes the architecture as built.
3
Manufacturer Specification
Provided by the system manufacturer but not independently verified. Moderate quality.
4
Manufacturer Claim
Marketing or product description claims. Lowest quality. Cannot substitute for architectural evidence.

A classification based entirely on Tier 1 and Tier 2 evidence is definitive. A classification based on Tier 3 or Tier 4 evidence carries uncertainty — the architecture described may differ from the architecture as implemented.

What the Questions Produce


The four questions produce one of three classification results. The result depends on which combination of answers was obtained.

Tier 1

In-the-Loop

Motion is directly physics-derived, delivered through independent axes from a center-of-mass reference, with timing and magnitude characteristics consistent with the depicted vehicle events.

  • Question 1: Physics engine output, no transformation layer
  • Question 2: Independent axes, center-of-mass reference
  • Question 3: Motion consistent with vehicle event parameters
  • Question 4: Evidence supports all three answers
Tier 2

Surface-Level

Motion is present and physically delivered to the participant, but it is not causatively derived from the vehicle physics state at time of generation. The inner ear receives motion information that does not directly correspond to what the vehicle is doing.

  • Question 1: Motion present but transformed (fails), OR
  • Question 2: Axes mechanically coupled, wrong reference (fails)
  • Motion exists but does not represent vehicle physics
Tier 3

Out-of-the-Loop

No physics-derived motion reaches the participant at all. The simulation is entirely visual and auditory. The inner ear receives no motion signal during the simulation event.

  • No motion platform present, OR
  • Platform present but not active during simulation, OR
  • Motion present but does not reach the participant's position

Six Things Classification Is Not Based On


Classification is sometimes confused with other kinds of quality assessment. These are the most common misreadings.

Classification is based on how realistic it feels
Classification is based on whether the motion corresponds to the physics
Feeling is subjective and variable. The classification question is architectural: is the motion causatively derived? That is measurable independently of participant impression.
More motion always means a higher classification
The type of motion determines classification, not the amount
A large-amplitude hexapod system with washout filtering is Surface-Level. A smaller system with direct physics-derived motion and independent axes is In-the-Loop. Amplitude is not assessed.
Six degrees of freedom means In-the-Loop
Six DOF and six independent axes are different things
A hexapod achieves six DOF through coupled kinematics. An In-the-Loop system requires six independent axes. The number of DOF is not the criterion — independence is.
A high-quality visual system improves the classification
Visuals are not assessed in the classification
Classification is about motion architecture only. Screen resolution, field of view, frame rate, and visual quality have no bearing on whether motion meets the structural criteria.
In-the-Loop means the system is safe for all users
Classification describes structural properties, not medical suitability
In-the-Loop means the motion corresponds to vehicle physics. Questions about user suitability, contraindications, or safety screening for specific populations are separate and outside the scope of classification.
Classification is permanent once determined
Classification applies to a specific configuration at a specific time
If the hardware, software architecture, or motion control logic changes, the classification must be re-evaluated. A classification of version 1.0 does not extend to version 2.0.