2. Occupant sensing technologies

An occupant classification system (OCS) must answer two questions before it allows an airbag to deploy: is a person present, and what kind of person or object is it. The restraint control module (RCM) uses that answer to decide whether the frontal passenger airbag should fire at full force, fire at reduced force, or be suppressed entirely. Four sensing principles dominate production vehicles: pressure bladder, capacitive, strain gauge, and weight-based (load cell) seat sensors. Each converts a different physical quantity into an electrical signal, and each has distinct strengths, blind spots, and calibration requirements.

Pressure Bladder Sensing

A pressure bladder is a sealed, fluid- or air-filled pad placed under the seat cushion foam. When an occupant sits down, their mass compresses the bladder and raises the internal pressure. A differential pressure sensor referenced to atmosphere converts that pressure into a voltage or a PWM signal proportional to the load. Because the bladder integrates load over its whole area, it responds to distributed weight rather than to a single point, which makes it relatively tolerant of seating position.

The transfer function is approximately linear over the working range. A representative production bladder might output 0.5 V at no load and reach 4.5 V at roughly 100 kg distributed load, giving about 40 mV per kilogram. The RCM applies a classification threshold table:

The main weakness is drift. Temperature changes the fill pressure, and foam aging changes the mechanical coupling between occupant and bladder. Production systems therefore run an automatic zeroing routine at every ignition cycle: with the seat believed empty, the RCM stores the current sensor output as the new zero reference. If a bag of groceries is left on the seat during zeroing, the reference is corrupted and the seat may be misclassified for that drive cycle.

Capacitive Sensing

A capacitive OCS measures the change in capacitance between an electrode mat in the seat cushion and the vehicle ground. The human body is largely water and therefore has a high dielectric constant, so an occupant increases the measured capacitance far more than an equivalent mass of dry luggage. A typical sensing mat is driven with an AC excitation of roughly 100 kHz, and the front-end converts the capacitance change into a digital count.

The key advantage is material discrimination. A 20 kg suitcase and a 20 kg child produce similar bladder pressures but very different capacitance shifts, because the suitcase has low dielectric coupling. This lets the system distinguish a child restraint (CRS) from a heavy object, which is exactly the case FMVSS 208 and ECE R94 care about. The trade-off is sensitivity to moisture and to the seat heater mat, which must be shielded or time-multiplexed so it does not corrupt the measurement.

Strain Gauge and Weight-Based Sensors

Strain gauge and load-cell systems measure force directly at the seat mounting points. Four strain gauges bonded to the seat frame rails form a Wheatstone bridge; seat load bends the rail, changes gauge resistance, and unbalances the bridge. The output is a small differential voltage, typically 1–3 mV/V of excitation, which an instrumentation amplifier boosts before the ADC. Because the measurement is at the structural interface, it captures the true vertical load regardless of how the occupant is distributed across the cushion.

Weight-based classification is the most direct method: sum the four corner loads, subtract the known seat structure weight, and compare the remainder against thresholds. A representative calibration gives about 2.5 mV per kilogram per corner at 5 V excitation. The practical difficulty is that the seat belt and the floor pan also carry load paths, so a tightly buckled belt can unload the seat and make a heavy adult look lighter. Good designs cross-check the seat load against belt tension and seat position.

Signal Chain and Classification Logic

Regardless of the physical principle, the signal path is similar: sense element → analog front-end → ADC → filtering → feature extraction → threshold comparison → classification output to the RCM. The figure below shows how the four technologies feed a common classification stage and how the result drives the airbag decision.

Practical Calibration and Fault Handling

Calibration is where these systems earn or lose their accuracy. A typical end-of-line procedure loads the seat with known masses at defined positions and records the sensor response, then writes the slope and offset into non-volatile memory. In service, the most common failure modes are:

• Zero drift: the empty-seat reference shifts with temperature or foam set. Symptom: intermittent "passenger airbag off" with an adult seated. Fix: re-run the zeroing procedure on a level surface with the seat empty.
• Connector or harness fault: an open strain gauge bridge or a shorted bladder line. Symptom: a fixed out-of-range reading and a stored DTC. Fix: check resistance across the bridge (typically 350 Ω nominal) and inspect the seat connector.
• Moisture ingress in a capacitive mat: a spilled drink raises baseline capacitance. Symptom: the seat reads occupied when empty. Fix: dry the cushion and re-zero; replace the mat if the shift persists.
• Mechanical preload: a foreign object trapped under the seat rail biases the load cell. Symptom: a constant offset on one corner. Fix: remove the obstruction and recalibrate.

The choice among these technologies is ultimately a trade study. Bladder systems are inexpensive and robust but cannot see material. Capacitive systems discriminate occupants from objects but are sensitive to moisture and heaters. Strain gauge and load-cell systems give the most direct mass measurement but require careful mechanical integration and belt-load compensation. Most modern vehicles combine at least two principles so that the classification decision is based on independent physical evidence rather than a single sensor that can drift or be fooled.