4. Anthropomorphic Test Devices (ATDs)

An Anthropomorphic Test Device (ATD) is a calibrated, instrumented human surrogate whose mechanical response has been tuned to approximate the impact response of a living human occupant within a defined loading regime. In restraint performance testing, the ATD is simultaneously the measurement platform and the physical interface with the restraint system. Every belt path, airbag deployment, and seat structure interaction is ultimately judged through the ATD's measured channels. Understanding which dummy family to use, how its biofidelity was established, how it is calibrated, and how its channels map to injury criteria is therefore the foundation of any credible restraint test program.

Dummy Families and Their Application Domains

ATDs are organized into families by occupant size, crash direction, and regulatory domain. The most common families encountered in restraint development are:

• Frontal impact family: Hybrid III in 50th percentile male, 5th percentile female, and 95th percentile male sizes; THOR (Test Device for Human Occupant Restraint) as the advanced mid-size male ATD with improved chest and pelvis instrumentation.
• Side impact family: ES-2re (50th percentile male), ES-2 (small female equivalent), SID-IIs (5th percentile female), and WorldSID 50th percentile male, which provides a more human-like ribcage and arm position.
• Rear impact family: BioRID II and RID2, designed specifically for neck injury assessment under whiplash loading.
• Child restraint family: Q-series (Q0, Q1, Q1.5, Q3, Q6, Q10) replacing the older P-series, with Q10 representing a 10-year-old child.

Selection is driven by the test protocol. A FMVSS 208 frontal barrier test uses a Hybrid III 50th male in the driver position and a Hybrid III 5th female in the passenger position. A Euro NCAP side pole test uses WorldSID 50th. A UN R129 child restraint test uses the Q-series appropriate to the child's mass group. Using the wrong family invalidates the comparison to the regulatory or consumer rating threshold, even if the physical test is otherwise well executed.

Biofidelity: What It Means and How It Is Judged

Biofidelity is the degree to which an ATD reproduces the biomechanical response of a human cadaver or volunteer under matched loading. It is not a single number; it is a profile of responses across body regions and loading rates. The standard evaluation framework is the Biofidelity Ranking System (BRS), which compares ATD response against cadaveric corridors for force–deflection, moment–rotation, and acceleration–time histories. A BRS score below 1.0 is considered excellent, 1.0–2.0 acceptable, and above 2.0 poor for a given body region.

Key biofidelity limitations are well documented. Hybrid III has a stiff thoracic spine and a simplified lumbar spine, which makes it a poor predictor of rear-impact neck injury. Its abdomen is essentially a rigid block, so it cannot assess submarining-related abdominal loading. WorldSID improved ribcage biofidelity and added a more human-like shoulder, but its arm position must be set carefully because it affects side airbag interaction. THOR introduced a more human-like thorax with four ribs and a realistic pelvis, but its complexity increases calibration burden and channel count.

Calibration: The Prerequisite for Valid Data

An ATD is only a measurement instrument after it has been calibrated. Calibration is performed at defined intervals (typically every test or every 5–10 tests, depending on the channel and the laboratory's quality system) and after any event that may have altered the dummy's response, such as a hard impact or a dropped head.

Calibration is channel-specific and uses dedicated fixtures. Representative procedures include:

• Head drop: The head is dropped onto a rigid flat plate from a specified height. The peak resultant acceleration at the head CG must fall within a corridor (for Hybrid III 50th male, approximately 225–275 g at 15 in drop height, with a specific pulse shape).
• Neck pendulum: The head–neck assembly is mounted in a pendulum and swung so the head experiences a controlled flexion or extension moment. The moment–angle response must lie within the certification corridor.
• Chest calibration: A pendulum impacts the chest at a defined velocity and mass. The chest deflection and the phase lag between force and deflection are checked.
• Knee and pelvis: Impact tests verify force–deflection characteristics of the knee slider and the pelvis iliac crest.

Each calibration produces a pass/fail result against a corridor, not a single value. A channel that fails calibration must be repaired or the dummy replaced before the test data can be considered valid. Calibration records are part of the test documentation and are frequently audited by regulatory bodies.

Channel Assignment and Injury Criteria

Channel assignment is the mapping of physical transducers to the injury criteria required by the test protocol. This mapping is not arbitrary; each criterion has a defined measurement location, axis, and filtering requirement. The table below summarizes the most common frontal and side impact channels.

Channel assignment also determines the data acquisition configuration. Each channel requires a specified sample rate (typically 10–20 kHz for accelerometers, 10 kHz for load cells), an anti-aliasing filter, and a post-test digital filter per SAE J211. The sign convention must be consistent: for a Hybrid III, +X is forward, +Y is to the dummy's left, +Z is upward, and moments follow the right-hand rule. Mixing sign conventions between channels is a common source of erroneous injury criteria.

Practical Channel Assignment Workflow

In a real restraint test, channel assignment begins with the test protocol, not with the dummy. The protocol defines which injury criteria must be reported. From those criteria, the required channels are derived, and from the channels, the transducer locations and data acquisition settings are determined. A typical workflow for a frontal sled test is:

1. Read the protocol and list every required injury criterion (e.g., HIC15, Nij, chest deflection, femur force).
2. Map each criterion to its ATD channel and confirm the dummy family supports that channel (e.g., Hybrid III 50th has no abdomen channel; THOR does).
3. Verify calibration status of each channel against the laboratory's calibration schedule and the manufacturer's corridor.
4. Configure the DAQ: sample rate, anti-aliasing filter, and SAE J211 digital filter class (CFC 1000 for accelerometers, CFC 600 for neck moments, CFC 180 for chest deflection).
5. Perform a pre-test channel check: zero offsets, shunt calibration, and a tap test to confirm signal integrity.
6. After the test, verify that no channel clipped or dropped out, and that the sign convention matches the expected loading direction.

Example: Channel Assignment for a 50th Percentile Male Frontal Sled Test

Consider a sled test at 40 km/h with a 50th percentile Hybrid III. The protocol requires HIC15, Nij, chest deflection, and femur force. The channel list would include: head CG triaxial accelerometer (CFC 1000), upper neck six-axis load cell (CFC 600 for moments, CFC 1000 for forces), chest deflection potentiometer (CFC 180), chest triaxial accelerometer (CFC 1000), and left/right femur load cells (CFC 600). The DAQ would be configured at 20 kHz for accelerometers and 10 kHz for load cells, with anti-aliasing at 1/3 of the sample rate. Pre-test calibration records for each channel must be attached to the test report. If the chest deflection channel fails its calibration corridor, the test cannot proceed until the chest is repaired or replaced, because the chest deflection criterion is a primary pass/fail metric.

Mastering ATD families, biofidelity limits, calibration discipline, and channel assignment is what separates a test that produces numbers from a test that produces defensible restraint performance data. The dummy is not a passive passenger; it is the instrument, and its configuration determines what the test can and cannot measure.