DISCRIMINATE INPUT SYSTEM FOR DECISION ALGORITHM

• US 20070100528A1
• Filed: 11/02/2005
• Published: 05/03/2007
• Est. Priority Date: 11/02/2005
• Status: Active Grant

• Current Assignee: Lear Corporation
• Original Assignee: Lear Corporation
• Inventors: Kennedy, Karl, Ricard, Christopher, Nathan, John
• Granted Patent: US 7,519,461 B2
• Time in Patent Office: Days
• US Class Current: 701/45
• CPC Class Codes: B60R 21/01516 using force or pressure sensing means
• CPC Class Codes: B60R 21/0155 sensing belt tension

Abstract

A method for discriminating input to a sensing algorithm used with an occupancy classification system includes generating a series of sensor signals from a sensor array. The method further includes generating a sensor signal from a belt tension sensor and generating a pattern recognition algorithm as a function of: the series of sensor signals from the sensor array and the sensor signal from the belt tension sensor.

Claims

1. A method for discriminating input to a sensing algorithm used with an occupancy classification system, said method including the steps of:

generating a series of sensor signals from a sensor array;

generating a sensor signal from a belt tension sensor; and

generating a pattern recognition algorithm as a function of said series of sensor signals from said sensor array and said sensor signal from said belt tension sensor.

2. The method of claim 1 further comprising pre-processing said series of sensor signals from said sensor array and said sensor signal from said belt tension sensor prior to generation of said pattern recognition algorithm.

3. The method of claim 2, wherein pre-processing comprises adjusting sensor offsets for generating signals workable within software limitations.

4. The method of claim 1 further comprising post-processing output from said pattern recognition algorithm.

5. The method of claim 4, wherein said post-processing comprises filtering said output.

6. The method of claim 4, wherein said post-processing comprises overruling a non-matching output relative to a series of matching outputs.

7. The method of claim 1, wherein generating said sensor signal from said belt tension sensor comprises generating at least one of a high tension signal indicating a seatbelt is tightly applied to an occupant or and a low tension signal indicating that said seatbelt is normally applied to said occupant.

8. The method of claim 7, wherein said high tension signal indicates a child safety seat is buckled by said seatbelt.

9. The method of claim 1, wherein generating said pattern recognition algorithm comprises analyzing said series of sensor signals from said sensor array and said sensor signal from said belt tension sensor in a neural net.

10. The method of claim 1, wherein generating said pattern recognition algorithm comprises ignoring said series of sensor signals from said sensor array in response to said sensor signal from said belt tension sensor indicating tension corresponding with a safety seat.

11. The method of claim 10, further comprising generating said pattern recognition in a neural net, whereby said step of ignoring comprises overriding output of said neural net and providing a predetermined output as said pattern recognition.

12. A method for discriminating input to a sensing algorithm used with the output of a sensor array and a seatbelt tension sensor for a vehicle seat occupancy sensing system used with a neural net for occupancy classification, said method including the steps of:

buckling at least one of a passenger and a safety seat into a particular vehicle seat with a seatbelt;

generating a series of sensor response signals from the sensor array and the belt tension sensor in response to pressure on said vehicle seat and tension on said seatbelt;

comparing each sensor response signal through the neural net; and

generating a classification signal in said neural net as a function of said response signals indicating that said tension on said seatbelt exceeds a predetermined threshold, and therefore said safety seat occupies said vehicle seat.

13. The method of claim 12 further comprising pre-processing said series of sensor response signals from said sensor array and said belt tension sensor prior to comparing each sensor response signal through the neural net.

14. The method of claim 13, wherein pre-processing comprises adjusting sensor offsets for generating signals workable within software limitations.

15. The method of claim 12 further comprising post-processing said classification signal.

16. The method of claim 15, wherein said post-processing comprises overruling a non-matching output relative to a series of matching outputs.

17. The method of claim 12, wherein generating said pattern recognition in said neural net comprises overriding output of said neural net and providing a predetermined output as said classification signal.

18. An occupant classification system for a vehicle having a vehicle seat having vehicle occupant sensors associated therewith, a seatbelt for belting in a seat occupant having a belt tension sensor measuring tension thereof, said occupant classification system comprising:

a controller comprising logic generating a series of sensor signals from the occupant sensors;

generating a sensor signal from the belt tension sensor; and

generating a pattern recognition algorithm as a function of said series of sensor signals from the occupant sensors and said sensor signal from the belt tension sensor.

19. The system of claim 18, wherein generating said pattern recognition algorithm comprises analyzing said series of sensor signals from the vehicle occupant sensors and said sensor signal from the belt tension sensor in a neural net.

20. The system of claim 18, wherein generating said pattern recognition algorithm comprises ignoring said series of sensor signals from the vehicle occupant sensors in response to said sensor signal from said belt tension sensor indicating tension corresponding to that typically encountered with a safety seat in the vehicle seat.