2. Occupant classification and OOP definitions

FMVSS 208 does not treat "the occupant" as a single entity. The standard defines a set of discrete occupant surrogates, seating configurations, and out-of-position (OOP) test positions, and it requires the airbag system to distinguish among them before deciding whether, when, and at what energy to deploy. This chapter explains how those definitions are constructed, how the classification decision is made in practice, and how suppression and low-risk deployment (LRD) logic are structured to satisfy the regulation without compromising the protection of normally seated occupants.

Why classification is a regulatory requirement, not a feature

FMVSS 208, as amended by the advanced airbag rulemaking, requires that the frontal airbag system either suppress deployment or deploy at reduced output when a small occupant or a child in a rear-facing child restraint is present in the right front passenger seat. The regulation expresses this through a set of test procedures using specific anthropomorphic test devices (ATDs) and specific positions. The engineering consequence is that the occupant sensing system (OSS) must produce a classification that is stable, repeatable, and defensible across the full matrix of ATDs, seat positions, and postures defined in the standard.

The four occupant categories that dominate the OOP and suppression test matrix are:

• 1-year-old (CRABI 12-month) — used primarily in rear-facing child restraint and OOP scenarios; the smallest surrogate in the matrix.
• 3-year-old (CRABI 3-year or Hybrid III 3-year-old) — mid-size child surrogate, frequently used for OOP near the instrument panel.
• 6-year-old (Hybrid III 6-year-old) — larger child surrogate, used to verify that the system does not misclassify a child as an adult.
• 5th percentile female (Hybrid III 5F) — the smallest adult surrogate; must be protected, not suppressed, in the normal seating position.

Each surrogate has a defined mass, seated height, and hip-to-knee geometry that directly affects the sensor signature the OSS observes. The 5F, for example, sits far forward relative to a 50th percentile male, which changes both the seat load distribution and the occupant-to-instrument-panel distance.

Seating configurations and the position matrix

FMVSS 208 defines the seating configuration in terms of seat track position, seat back angle, and head restraint adjustment. The standard specifies a "mid-track" reference and a range of fore-aft positions. For OOP testing, the seat is typically placed at the mid-track or full-forward position, and the ATD is placed in one of several defined postures relative to the instrument panel and the deploying airbag module.

The OOP positions are not arbitrary. They are constructed so that the airbag, during deployment, contacts the ATD in a way that represents a realistic misuse or crash-induced posture. The principal OOP positions include:

• Position 1 / Position 2 — child restraint or child ATD positioned close to the instrument panel, with the head or torso in the airbag deployment path.
• Position 3 — child ATD leaning forward, head near the top of the instrument panel.
• Position 4 / Position 5 — child ATD out of position laterally or with the head turned, so the airbag contacts the side of the head or neck.
• Position 6 — 5F or child ATD with the torso against the instrument panel, representing a "brace" posture.

Each position is defined by measurable coordinates: the horizontal distance from the ATD's nose or chest to the instrument panel, the vertical height of the head, and the orientation of the ATD relative to the vehicle centerline. These coordinates are what the test laboratory sets up, and they are also what the OSS must implicitly resolve from sensor data.

From sensor signals to a classification decision

A typical passenger-side OSS combines a seat weight sensor (or bladder pressure sensor), a seat belt tension/usage sensor, and sometimes a capacitive or radar-based occupant presence sensor. The raw signals are conditioned and fused into a feature vector, and a classification algorithm maps that vector to one of the regulated categories: adult (full deployment), small adult / 5F (full or reduced deployment depending on the specific requirement), child (suppress or LRD), or empty seat (suppress).

The classification is not a single threshold. Consider a simplified weight-based example. Let W be the filtered seat load in kilograms, and let the system use two thresholds, T_child and T_adult:

if W < T_child:        classify = EMPTY_OR_CHILD   -> suppress
elif W < T_adult:      classify = CHILD             -> suppress or LRD
else:                  classify = ADULT             -> full deployment

In a real system, T_child and T_adult are not fixed constants. They are calibrated per seat and per vehicle, and they are adjusted for seat belt tension, because a tightly belted child restraint can load the seat differently than an unbelted child. A representative calibration might place T_child near 20 kg and T_adult near 45 kg, but these numbers are illustrative only and must be derived from the actual seat structure and sensor response.

The critical engineering problem is the overlap region. A 6-year-old ATD and a 5F ATD can produce similar seat loads depending on posture and belt routing. The classifier must therefore use additional features — belt tension, occupant height proxy from a capacitive sensor, or seat back pressure distribution — to separate them. This is why single-sensor weight-only systems are insufficient for FMVSS 208 compliance in modern vehicles.

Suppression and inhibition logic

Suppression logic is the decision layer that converts the classification into an airbag command. The logic must satisfy three constraints simultaneously:

1. Suppress when required — child restraint and small child in the OOP positions must not receive a full-energy deployment.
2. Do not suppress when protection is required — the 5F in the normal seating position must receive deployment; suppressing her would fail the injury criteria.
3. Fail safe — a sensor fault must not produce an unsafe deployment decision. The typical fail-safe behavior is to default to full deployment for the driver and to a calibrated default for the passenger, with a telltale warning to the occupant.

The logic is usually implemented as a state machine with hysteresis to prevent chatter. For example, the system may require the classification to be stable for a defined debounce time (e.g., 200–500 ms of continuous sensor agreement) before changing the deployment command. This prevents a transient load — such as an occupant shifting in the seat — from toggling suppression on and off.

Calibration and verification in practice

Calibrating the classifier is an empirical exercise. The seat is instrumented, each ATD is placed in each defined position, and the sensor outputs are recorded across the expected environmental range (temperature, seat cushion compression, belt routing). The recorded data are used to set the thresholds and to verify that the decision boundaries separate the required categories with margin.

A practical verification sequence for a passenger-side OSS looks like this:

1. Install the seat and OSS in the vehicle buck at the mid-track position.
2. Place the 1-year-old ATD in a rear-facing CRS; confirm the system reports suppression and that the telltale indicates passenger airbag off.
3. Place the 3-year-old and 6-year-old ATDs in each OOP position; confirm suppression or LRD as required, and record the sensor values at the decision boundary.
4. Place the 5F ATD in the normal seating position; confirm full deployment command and that the telltale indicates passenger airbag on.
5. Introduce a sensor fault (open circuit or out-of-range signal); confirm the fail-safe default and the telltale warning.

Typical failure symptoms during this sequence are instructive. If the 6-year-old is classified as an adult, the seat load threshold T_adult is too low or the belt-tension feature is not being weighted correctly. If the 5F is suppressed, T_adult is too high or the debounce time is masking a valid adult classification. If the classification chatters between states, the hysteresis band is too narrow or the debounce window is too short. Each symptom points to a specific calibration parameter rather than a generic "sensor problem."

Suppression logic and the telltale requirement

When the system suppresses the passenger airbag, FMVSS 208 requires a telltale to inform the occupant. The telltale logic is part of the suppression state machine: it must illuminate when suppression is active and extinguish when full deployment is restored. The telltale is not a diagnostic afterthought; it is a regulated output, and its timing and reliability are verified alongside the deployment command.

The complete decision chain — sensor acquisition, fusion, classification, suppression command, and telltale — is what the regulation actually tests. The ATD positions and seating configurations define the input space; the suppression and inhibition logic define the output space; and the calibration and verification activities connect the two. Understanding this chain is the prerequisite for the OOP injury-mitigation design work that follows.