Protecting military perimeters from approaching human and vehicle using biologically realistic neural network

• US 8,615,476 B2
• Filed: 04/13/2010
• Issued: 12/24/2013
• Est. Priority Date: 04/15/2009
• Status: Active Grant

• Current Assignee: University of Southern California
• Original Assignee: University of Southern California
• Inventors: Berger, Theodore W., Dibazar, Alireza, Park, Hyung O.
• Primary Examiner(s): VINCENT, DAVID ROBERT
• Application Number:
  US12/759,556
• Publication Number:
  US 20100268671A1
• Time in Patent Office: 1,351 Days
• Field of Search: 706/12, 706 15- 42, 706/45, 706/62
• US Class Current: 706/12
• CPC Class Codes: G06N 3/049 Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
• CPC Class Codes: G08B 13/1663 using seismic sensing means

Abstract

An approaching human threat or vehicle, such as a suicide bomber nearing a secured zone such as a military base, may be detected and classified. A vibration recognition system may detect a systematic vibration event. The entity might be a medium, human, animal, or a passenger vehicle. The system may discriminate between such an event and a background or other vibration event, such as a falling tree limb. A seismic sensor may be employed to detect vibrations generated by footsteps and a vehicle. Seismic waves may be processed locally where the sensor is located. The system may wirelessly communicate with a remote command center. Temporal features of the vibration signals may be modeled by a biologically realistic neural network with good false recognition rates. The models may reject quadrupedal animal footsteps.

Claims

1. A security system for protecting perimeters from an approaching human or vehicle comprising:

a seismic sensor system configured to sense ground vibration, including ground vibration caused by each of the following;

human footsteps, a vehicle, and quadrupedal animal footsteps; and

a trained biologically realistic neural network system configured to receive information indicative of the ground vibration sensed by the sensor system and to discriminate between different causes of the ground vibration based on this information, including between ground vibration caused by each of the following;

human footsteps;

a vehicle; and

quadrupedal animal footsteps,

wherein the trained biologically realistic neural network system includes a plurality of trained biologically realistic neural networks in a parallel configuration.

2. The security system of claim 1 wherein the trained biologically realistic neural network system is a trained biologically realistic dynamic neural network system.

3. The security system of claim 1 wherein each of the plurality of trained biologically realistic neural networks is configured to receive the information indicative of the ground vibration sensed by the sensor system.

4. The security system of claim 3 wherein the plurality of trained biologically realistic neural networks includes:

a first trained biologically realistic neural network trained to discriminate ground vibration caused by human footsteps;

a second trained biologically realistic neural network trained to discriminate ground vibration caused by a vehicle;

a third trained biologically realistic neural network trained to discriminate ground vibration caused by quadrupedal animal footsteps; and

a forth trained biologically realistic neural network trained to discriminate ground vibration caused by other activity.

5. The security system of claim 4 wherein the trained biologically realistic neural network system is configured to designate the ground vibration as being caused by the type of vibration which is being discriminated by the trained biologically realistic neural network which indicates the strongest degree of correlation with the ground vibration.

6. The security system of claim 5 wherein the strongest degree of correlation is indicated by the trained biologically realistic neural network which has the lowest firing rate.

7. The security system of claim 5 wherein lateral inhibition is used as part of the discrimination process.

8. The security system of claim 1 wherein each trained biologically realistic neural network is trained to predict the input.

9. The security system of claim 8 wherein the output of each trained biologically realistic neural network is fed back to the input.

10. The security system of claim 9 wherein each trained biologically realistic neural network is configured to substantially implement the following algorithm:

O(t,Θ

)=y(t)−

ŷ

(t,Θ

)

where, Θ

is the vector of all the weights and thresholds of the network and y is the current value of the input, hat of y is estimated input, and O is the output of each trained biologically realistic neural network.

11. The security system of claim 9 wherein the input output relationship of each trained biologically realistic neural network substantially conforms to the following equation:

12. The security system of claim 1 wherein:

the information indicative of the ground vibration sensed by the sensor system includes temporal patterns of the vibration; and

wherein the trained biologically realistic neural network system is configured to examine the temporal patterns.

13. The security system of claim 1 wherein:

the seismic sensor system includes a plurality of seismic sensors; and

the trained biologically realistic neural network system is trained to determine the location of the ground vibration sensed by the plurality of seismic sensors.

14. The security system of claim 13 wherein the trained biologically realistic neural network system is trained to determine the location of the ground vibration based on the plurality of sensors being arranged in a grid-like pattern.

15. The security system of claim 14 wherein the trained biologically realistic neural network system includes a trained biologically realistic neural network system associated with each of the seismic sensors.

19. The security system of claim 1 wherein the trained biologically realistic neural network system is recurrent and discrete.

16. A security system for protecting perimeters from an approaching human or vehicle comprising:

a seismic sensor system configured to sense ground vibration, including ground vibration caused by each of the following;

human footsteps and a vehicle; and

a trained biologically realistic neural network system configured to receive information indicative of the ground vibration sensed by the sensor system and to discriminate between different causes of the ground vibration based on this information, including between ground vibration caused by each of the following;

human footsteps and a vehicle,

wherein the trained biologically realistic neural network system includes a plurality of trained biologically realistic neural networks in a parallel configuration.

17. The security system of claim 16 wherein each of the plurality of trained biologically realistic neural networks is configured to receive the information indicative of the ground vibration sensed by the sensor system.

18. The security system of claim 17 wherein the plurality of trained biologically realistic neural networks includes:

a first trained biologically realistic neural network trained to discriminate ground vibration caused by human footsteps;

a second trained biologically realistic neural network trained to discriminate ground vibration caused by a vehicle; and

a third trained biologically realistic neural network trained to discriminate ground vibration caused by the other activity.

20. The security system of claim 16 wherein the trained biologically realistic neural network system is recurrent and discrete.