Neural Network Systems for Vehicles

• US 20080144944A1
• Filed: 02/28/2008
• Published: 06/19/2008
• Est. Priority Date: 05/05/1992
• Status: Active Grant

• Current Assignee: American Vehicular Sciences LLC (Acacia Research Corporation)
• Original Assignee: Automotive Technologies International Incorporated
• Inventors: Breed, David S.
• Granted Patent: US 7,660,437 B2
• Time in Patent Office: Days
• US Class Current: 382/224
• CPC Class Codes: G06V 20/59 inside of a vehicle, e.g. relating to seat occupancy, driver state or inner lighting conditions
• CPC Class Codes: G06V 40/103 Static body considered as a whole, e.g. static pedestrian or occupant recognition

Abstract

Method for obtaining information about an occupying item in a space in a vehicle in accordance with the invention includes obtaining images of an area above a seat in the vehicle in which the occupying item is situated and classifying the occupying item by inputting signals derived from the images into a trained neural network form which is trained to output an indication of the class of occupying item from one of a predetermined number of possible classes. The method is applicable for various vehicles including automobiles, trucks, buses, airplanes and boats. The images may be pre-processed to remove background portions of the images and then converted into signals for input into the neural network form.

Claims

1. A method for obtaining information about an occupying item in a space in a vehicle, comprising:

obtaining images of an area above a seat in the vehicle in which the occupying item is situated; and

classifying the occupying item by inputting signals derived from the images into a trained neural network form which is trained to output an indication of the class of occupying item from one of a predetermined number of possible classes.

2. The method of claim 1, further comprising:

pre-processing the images to remove background portions of the images; and

converting the processed images into signals for input into the neural network form.

3. The method of claim 2, wherein the step of pre-processing the images comprises performing edge analysis to identify a region of interest of the occupying item and edges of the region of interest of the occupying item and then removing portions of the images outside of the edges of the region of interest of the occupying item.

4. The method of claim 1, further comprising:

analyzing multiple images obtained from the same imager;

pre-processing the images by removing portions of the images which are the same in multiple images; and

converting the processed images into signals for input into the neural network form.

5. The method of claim 1, wherein the neural network form is a modular neural network.

6. The method of claim 5, wherein at least one of the modular neural networks includes a plurality of submodular neural networks.

7. The method of claim 5, wherein the vehicle includes a seat on which the occupying item may be present, the step of classifying the occupying item comprises training each of the modular neural networks to determine whether the occupying item is one of a predetermined number of known occupying items.

8. The method of claim 1, wherein the vehicle includes a seat on which the occupying item is present, the step of classifying the occupying item comprises:

forming different subsystems of the neural network form;

directing images to one of the different subsystems as a function of light conditions; and

in each subsystem, providing a first neural network to determine whether there is an occupying item on the seat, providing a second neural network to determine whether there is an infant or child seat on the seat and providing a third neural network to determine whether there is an adult on the seat.

9. The method of claim 1, wherein the neural network form is a combination neural network including a plurality of neural networks, each arranged to classify the occupying item using different input or to provide a different output.

10. The method of claim 1, wherein the vehicle is a selected from a group consisting of an automobile, a truck, a bus, an airplane and a boat.

11. The method of claim 1, wherein the occupying item is a human occupant, further comprising controlling actuation of an occupant protection or restraint device based on the classification of the occupant.

12. The method of claim 1, wherein the step of classifying the occupying item comprises determining whether the occupying item is identifiable as one of a number of predetermined possible occupying items.

13. A method of classifying a seat occupant or a state of the seat occupant into one of a first category and a second category in a vehicle having an array of sensors and a controller connected to the sensors for embodying the method, comprising:

obtaining respective sensor signals from the sensors;

inputting the obtained sensor signals into a neural network form trained to provide a first or second value as a decision about a classification of the seat occupant; and

determining the first or second value via the neural network form.

14. The method of claim 13, wherein the sensors are optical sensors arranged to obtain images of the seat occupant.

15. The method of claim 14, further comprising:

pre-processing the images to remove background portions of the images; and

converting the processed images into signals for input into the neural network form.

16. The method of claim 14, further comprising:

forming different subsystems of the neural network form;

directing sensor signals derived from the images to one of the different subsystems as a function of light conditions; and

in each subsystem, providing a first neural network to determine whether there is an occupant on the seat, providing a second neural network to determine whether there is an infant or child seat on the seat and providing a third neural network to determine whether there is an adult on the seat.

17. The method of claim 13, wherein the neural network form is a modular neural network.

18. The method of claim 17, wherein at least one of the modular neural networks includes a plurality of submodular neural networks.

19. The method of claim 17, further comprises training each of the modular neural networks to determine whether the occupant is one of a predetermined number of known occupant types and output either the first value or the second value depending on whether the occupant is of the occupant type for which the modular neural network is trained.

20. The method of claim 13, wherein the neural network form is a combination neural network including a plurality of neural networks, each arranged to classify the occupying item using different input or to provide a different output.

21. The method of claim 13, further comprising controlling actuation of an occupant protection or restraint device based on the output of the first or second value from the neural network form.