Pattern recognition techniques, such as neural networks, have been appiied to identify objects within the passenger compartment of the vehicle, such as a rear facing child seat or an out-of-position occupant, and to suppressth e airbagw hen an occupanti s more likely to be injured by the air-bag than by the accident. Neural networks have also been applied to sense automobile crashes. The use of neural networks is extended here to tailoring the airbag inflation to the severity of the crash, the size, position and relative velocity of the occupant and other factors such as seatbelt usage, seat and seat back positions, vehicle velocity, and any other relevant information.

It is well known that a neural network based crash sensor can forecast, based on the first part of the crash pulse, that the crash wilI be of a severity which requires that an airbag be deployed. This is extended here to enhance the capabilities of this sensor to forecast the velocity change of the crash over the entire crash period. Then a pattern recognition occupant position and velocity determination sensor is added. Finaiy, an occupant weight sensor is inctuded to permit a measure of the occupant’s momentum or kinetic energy. The combination of these systems in various forms will be used to optimize inflation and/or deflation of the airbag to create a “smart airbag” system.

Crash sensors can predict that a crash is of a severity which requires the deployment of an airbag for the majority of real world crashes. A more difficult problem is to predict the crash velocity versus time function and then to adjust the airbag inflation/deflation over time so that just the proper amount of gas is in the airbag at all times even without considering the influence of the occupant. To also simultaneously consider the occupant size, weight, position and velocity renders this problem unsolvable by conventional methods.