Airbags are crucial elements of passive safety in vehicles that help minimizing occupant injuries during various crash scenarios such as frontal, side, and oblique impacts. Airbags in cars are now mandatory in many countries, and their performance depends on how well the system is designed. A well-tuned airbag deployment algorithm is necessary to score superior NCAP safety ratings. Tuning of airbag deployment algorithms requires several data points which are obtained through actual crash testing. This is a cumbersome and expensive process as it involves crash tests for each scenario (e.g., full front barrier, offset deformable barrier, angled impact, etc.) at multiple test speeds. These tests are destructive and render the vehicles only worthy of scrap. The data gathered from various sensors (acceleration, pressure, etc.) is used to develop robust vehicle model specific algorithms that must correctly identify the crash scenario and send airbag firing signal at the optimal pre-decided time. The question is, can we reduce the number of crash tests and still develop an equally robust airbag deployment system?
In this work, we discuss a novel method of crash pulse sensing. It is a semi-empirical model that uses both physics-based and data-based approach. Our proposed model uses logged crash data for certain speeds and generates crash pulse for the remaining required speeds - thus reducing the number of required actual tests. Model is validated for different crash scenarios. We estimate that this model could potentially reduce a significant cost and development time.