[   Home   ] [ Services ] [ Gallery  ] [ About Us ] [ Contact  ] [  Links   ] [ Site Info ]
HORK Enterprises
3221 Quick Road
Holly, MI 48442, USA
 +1 (248) 328-0231

Occupant Spacial Sensor System

The Occupant Spatial Sensor (OSS) System is designed to classify the occupant of a vehicle and detect his/her position relative to the airbag. This allows the airbag control module to be programmed to deploy the airbag with a power that is appropriate for the situation at hand. The technology used for the Occupant Spatial Sensor has been developed by Dr. David S. Breed and Automotive Technologies International, Inc. since before 1992. Numerous patents, such as
US Patent 6,529,809
and others contained therein or still pending, cover the ideas underlying the technology.
[ Echoes ] In its first incarnation, the OSS placed four ultrasonic transducers either around the passenger or the driver seating area. These ultrasonic transducers are connected to an electronic control module that excites them at their natural frequency to send an inaudible sound pulse into the designated seating area. Echoes bouncing of occupants or objects in the seat are received back by the transducers and the control module. There the signals are massaged, digitized, and combined into a time discretized series of numbers, known as a pattern or "vector".
Pattern Recognition
[ ANN ] The second part of the Occupant Spacial Sensor System is the pattern recognition software that performes the occupant classification and determines whether the occupant is at risk of injury from an airbag deployment. That would be the case if the occupant is in fact a rearward facing infant or if the occupant is in close proximity to the airbag module, a.k.a. Out Of Position (OOP).
For this pattern recognition, the system uses an Artificial Neural Network (ANN), that is particularly well suited for such a task. It is a special piece of software that obtains its parameters from being trained on patterns it needs to recognize. For this a large set of vectors is collected from a variety of occupants in both "at risk" and normal seating positions. The network training consists of feeding these vectors and the desired outcome through the ANN. During this "learning" process the software optimizes its parameters, such that similar patterns will also result in the desired decision.
The system was jointly developed by ATI and Autoliv and was first implemented in the Model Year 2001 Jaguar XK8.

Henk was fortunate enough to be David's "Chief Engineer" on this program, until it was taken over by Autoliv engineers for production, working with a very talented ATI team comprising:
David Breed, Tie-Qi Chen, Will DuVall, Pete Johnson, Jerry McCleave, Greg McCormick, Jeff Morin, Chrissy Pikulas, Stephen Smith, Dan Thomas, Mike Timek, Jeff Weber, Krista Xu, Lilian Ying, and the many patient occupants that contorted themselves into all kinds of positions, to generate the patterns needed for ANN training.