Pedestrian head injury prediction based on vehicle section image using machine learning.

Abstract

Child occupant injury performance evaluation is integral part of most Consumer based and Insurance based vehicle safety evaluation protocols worldwide. New CAR Assessment Protocols (NCAP) now have separate ratings exclusively to evaluate Child dummy performance for different test scenarios like Frontal ODB, Full frontal, Side MDB and Side Pole etc. Hence all vehicle equipment manufacturers have need and focus of maximizing child injury performance in vehicles. Sled tests are proven method of optimizing various restraint systems like Seat belts, Driver Airbags, Passenger Airbags, Steering columns and Seats etc. Obvious benefits sled tests is that single fixture can be used for multiple tests, thereby avoiding the need of multiple vehicles in development stage which are quite expensive at early stages of vehicle development. For frontal type of impact cases, acceleration or deceleration based sled tests can be used depending on test facility available. In typical frontal impact scenarios like Frontal impact with rigid wall, Offset barrier impact or Pole impacts vehicle will be subjected to Pitching, Yawing and Rolling motion before bouncing back motion. Obviously, the motions are pronounced as we move front of vehicle to rear end. (from first row to second row and third row if any etc) Sled tests are conventionally used for tuning restraints in first row of occupants with good correlation, as effect of pitch, roll and yaw motions in first row are relatively less. However, as one move to second row, effect of pitching and rolling motions experienced by dummies are relatively high. Hence one should be careful of results from sled test. In the present study an attempt is made to simulate child injury performance of P3 dummy positioned on second row seat on defined child seat for 64 kmph frontal Offset deformed barrier type test. Sled pulses are carefully tuned to capture key injury patterns. Thence restraint parameters are tuned to improve child dummy injuries. In the last part of study correlated sled model will be used to compare P3 series child dummies vs Q3 series of child dummies. A comparative study of P3 vs Q3 dummies will be presented for a given vehicle pulse.