Vehicle crowd sensing system and method

1. A method of selective sensing for a vehicle crowd sensing system, comprising the steps of:

receiving a plurality of detection reports from a first set of vehicles, wherein each detection report includes an event, a position qualifier of the event, and a severity qualifier of the event;

developing an inherent error model for the event, wherein the inherent error model includes a weighted summation of a number of the detection reports, a standard deviation of the position qualifiers, and a standard deviation of the severity qualifiers

determining a resampling instruction for the event, wherein the resampling instruction is based on an action model and wherein the action model includes a proportional integral derivative policy evaluation of a result of the inherent error model; and

transmitting the resampling instruction for the event to a second set of vehicles.

2. The method of claim 1 , wherein the second set of vehicles are within a prescribed temporal or spatial range of the event.

3. The method of claim 1 , wherein the resampling instruction causes a first portion of the second set of vehicles to upload detection reports for the event.

4. The method of claim 3 , wherein the resampling instruction causes a second portion of the second set of vehicles to not upload detection reports for the event.

5. The method of claim 1 , wherein the weighted summation includes weights, and wherein the weights are calibratable values based on a classification of the event.

6. The method of claim 1 , wherein the result of the inherent error model provides a hybrid criteria value calculated in accordance with the following:

h

=

w

1

·

n

-

n

p

⁢

⁢

0

n

p

⁢

⁢

0

+

w

2

·

n

-

n

s

⁢

⁢

0

n

s

⁢

⁢

0

+

w

3

·

s

p

⁢

⁢

0

-

s

p

s

p

⁢

⁢

0

+

w

4

·

s

s

⁢

⁢

0

-

s

s

s

s

⁢

⁢

0

wherein h is the hybrid criteria value; w1, w2, w3, and w4 are the weights; n is the number of the detection reports; Sp is the standard deviation of the position qualifiers for the event; ss is the standard deviation of the severity qualifiers for the event; and np0, ns0, Sp0, and ss0 are reference targets.

7. The method of claim 6 , further comprising when the result of the inherent error model is less than a hybrid criteria threshold, the action model is disabled and the resampling instruction is set to 0%.

8. The method of claim 6 , further comprising when the result of the inherent error model is greater than a hybrid criteria threshold, the action model is enabled and the resampling instruction is set to a value between 0% and 100%.

9. The method of claim 1 , wherein the proportional/integral/derivative (PID) policy evaluation includes a proportion of the result of the inherent error model, an integral of the result of the inherent error model, and a derivative of the result of the inherent error model, each of the proportion, the integral, and the derivative being selectively usable for the action model.

10. The method of claim 1 , wherein each detection report further includes a set of time qualifiers of the event.

11. A method of selective sensing for a vehicle crowd sensing system, comprising the steps of:

receiving, at a vehicle, a resampling instruction from a backend facility for an event, wherein the resampling instruction is based on an action model and the action model is at least partly a factor of an inherent error model, wherein the inherent error model includes a weighted summation of a number of the detection reports, a standard deviation of the position qualifiers, and a standard deviation of the severity qualifiers;

determining, by the vehicle, whether the resampling instruction provides for uploading or non-uploading;

when the resampling instruction for the event provides for uploading, uploading, by the vehicle, a detection report for the event to the backend facility, wherein the detection report includes a position qualifier of the event and a severity qualifier of the event; and

when the resampling instruction for the event provides for non-uploading, not uploading, by the vehicle, the detection report for the event to the backend facility.

12. The method of claim 11 , wherein the resampling instruction is a probability, and the determining step uses the probability, a base sampling rate, a local cost, and a local confidence to determine whether the resampling instruction provides for uploading or non-uploading.

13. The method of claim 12 , wherein the base sampling rate, the local cost, and the local confidence are individual vehicles' parameters.

14. A vehicle crowd sensing system, comprising:

a server that includes a processor and non-transitory computer-readable memory, the non-transitory computer-readable memory storing a computer program; and

a database that is accessible by the server, the database storing a plurality of detection reports for an event and received from a first set of vehicles, wherein each detection report includes a position qualifier of the event and a severity qualifier of the event;

wherein the computer program, when executed by the processor, causes the server to:

develop an inherent error model for the event, wherein the inherent error model includes a weighted summation of a number of the detection reports, a standard deviation of the position qualifiers, and a standard deviation of the severity qualifiers,

determine a resampling instruction for the event, wherein the resampling instruction is based on an action model and wherein the action model includes a proportional integral derivative policy evaluation of a result of the inherent error model, and

transmit the resampling instruction for the event to a second set of vehicles.