This content is not included in your SAE MOBILUS subscription, or you are not logged in.

The Autonomous Recognition of Left Behind Passengers in Parked Vehicles

Journal Article
2011-01-0582
ISSN: 1946-3995, e-ISSN: 1946-4002
Published April 12, 2011 by SAE International in United States
The Autonomous Recognition of Left Behind Passengers in Parked Vehicles
Sector:
Citation: Fischer, C., Fischer, T., and Tibken, B., "The Autonomous Recognition of Left Behind Passengers in Parked Vehicles," SAE Int. J. Passeng. Cars – Mech. Syst. 4(1):509-522, 2011, https://doi.org/10.4271/2011-01-0582.
Language: English

Abstract:

The unattended left behind of children in parked vehicles is one of the major causes of lethal or serious injuries to children in non-traffic accidents in the U. S. For this reason Delphi is interested in the development of a low cost left behind occupant recognition so that its safety division launched the evaluation of different approaches for a reliable detection system in 2008. This contribution discusses the sensor evaluation, experiments under different conditions and the classification via data mining algorithms based on two potential approaches. The first one uses high sensitive analogue accelerometers at the car chassis and the second one is based on a pressure mat in the seat. Occupants inside the vehicle produce vibrations at the car chassis which can be monitored by the accelerometers. The needed electronic and different experimental results are explained in regard to an autonomous left behind recognition. Subsequently the experiments with the pressure mat are discussed. This system is appropriated for the determination of the physiological parameters of the seated occupant and is based on a reengineered series product - PODS-B. Its signal processing and the extracted signals are described and the relevance for the following classification paragraph becomes emphasized. Finally the gained information of the evaluated sensing elements is used to rate four different classification methods (k-NN, SVM, J48, PNN). The results are discussed and the proper classifiers are highlighted. Out of this the further development steps are defined and the benefit of the applications is characterized.