2. Anthropomorphic test devices for children

Child occupant protection scoring is only as credible as the measuring instrument inside the child restraint. That instrument is the anthropomorphic test device (ATD), commonly called a crash test dummy. In child protection assessment, three ATD families dominate: the Q-series, the Hybrid III child dummies, and CRABI. Each family embodies a different design philosophy, instrumentation set, calibration regime, and biofidelity envelope. Because scoring thresholds — HIC, chest deflection, neck loads, abdominal pressure — are tied to specific ATD designs, a score is meaningless unless the dummy family, its instrumentation, and its calibration state are known and documented.

Design philosophies and family characteristics

The Q-series (Q0, Q1, Q1.5, Q3, Q6, Q10) was developed primarily in Europe as a coherent scaling family, with the Q10 approaching the size of a small adult female. Q dummies are designed around child anthropometry and use a mix of accelerometers, load cells, and deflection sensors; the Q3 and Q6 are the workhorses of ECE R129 (i-Size) assessment. Their instrumentation is modular, and the family shares a common design language, which simplifies test setup across sizes.

The Hybrid III child dummies (typically the 3-year-old and 6-year-old, plus the CRABI-derived infant) descend from the adult Hybrid III architecture. They are widely used in FMVSS 213 and in many NCAP and research programs. Their heads, necks, and chests are built to a mature, well-characterized design, and their instrumentation is familiar to most test laboratories. The trade-off is that scaling an adult design down to a child is not biofidelic in every respect — particularly in the neck and abdomen.

CRABI (Child Restraint Air Bag Interaction) exists in 12-month and 18-month versions and was developed specifically to study infant interaction with deploying air bags. CRABI is not a general-purpose infant ATD for all restraint scoring; its instrumentation and biofidelity are optimized for the air-bag interaction problem, and its use in a scoring program must be justified by the specific injury mechanism being assessed.

Instrumentation: what is measured and where

Scoring metrics are computed from a small set of channels. The table below summarizes the typical instrumentation for a mid-size child ATD and the metric each channel feeds.

Two practical points matter for scoring. First, filtering: HIC and 3 ms clip values depend on the channel filter class (typically CFC 1000 for head acceleration, CFC 600 for chest). Applying the wrong filter changes the score. Second, sensor mounting: a head accelerometer mounted at the CG versus offset from it will read different values under rotation, so the mounting must match the ATD drawing and the scoring specification.

Calibration and its effect on score validity

Every ATD must be calibrated before a test series and re-verified at defined intervals. Calibration is not a formality: a drifted load cell or a stiffened neck changes the measured metric directly. The core calibration checks are:

• Head drop test — drop the head assembly from a specified height onto a rigid plate and verify the peak resultant acceleration and pulse shape against the ATD specification.
• Neck pendulum test — swing the head-neck assembly and verify the bending moment and angular displacement response.
• Chest calibration — apply a known deflection and verify the sensor output and return-to-zero behavior.
• Sensor zero and gain check — verify each channel against a shunt or known input before and after the test.

A common failure mode is a pre-test zero drift on the chest deflection channel: the sensor reads a small offset, the recorded compression is biased, and the chest score is wrong even though the test itself ran correctly. The troubleshooting step is to re-zero, re-shunt, and if drift persists, replace the sensor before testing. Similarly, a neck load cell that fails its pendulum check will produce a neck score that cannot be defended, regardless of how clean the acceleration traces look.

Biofidelity considerations in scoring

Biofidelity is the degree to which an ATD reproduces human response. It is size-, direction-, and metric-dependent. The Q-series was tuned with child-specific data and generally performs better in frontal and side child loading than scaled adult designs; the Hybrid III child dummies are well characterized but inherit adult-design compromises, especially in neck and abdomen response. CRABI is biofidelic for the infant air-bag interaction case but is not a general infant restraint ATD.

For scoring, biofidelity enters in three ways. First, threshold provenance: injury criteria such as HIC15 or chest compression limits were derived from specific ATD and PMHS data, so the threshold must match the ATD. Second, metric applicability: a metric validated on one ATD family may not transfer to another without re-derivation. Third, uncertainty: where biofidelity is weaker, the scoring program should widen tolerance bands or flag the result as indicative rather than definitive.

Putting it together: a defensible scoring workflow

A defensible score follows a traceable chain. Select the ATD family and size that match the injury mechanism and the regulatory or program requirement. Document the instrumentation channels and filter classes. Verify calibration immediately before the test series, and record zero and shunt checks. Run the test, then post-process with the specified filters and compute metrics. Finally, compare each metric against a threshold whose provenance matches the ATD. If any link is missing — wrong ATD for the threshold, unverified calibration, wrong filter — the score should be flagged rather than reported as a clean pass or fail.

In practice, the most frequent scoring errors are not in the crash itself but in the metadata: a Q3 threshold applied to a Hybrid III 3-year-old, a chest deflection channel left un-zeroed, or a head acceleration channel filtered at the wrong class. Treating the ATD, its instrumentation, and its calibration as part of the score — not as background setup — is what separates a defensible child protection assessment from a number that merely looks precise.