Systems and methods involving features of adaptive and/or autonomous traffic control

• US 10,699,561 B2
• Filed: 09/11/2017
• Issued: 06/30/2020
• Est. Priority Date: 11/22/2011
• Status: Active Grant

• Current Assignee: Fastec International, LLC
• Original Assignee: Fastec International, LLC
• Inventors: Robinson, Kurt B.
• Primary Examiner(s): Khatib, Rami
• Application Number:
  US15/700,876
• Publication Number:
  US 20180096594A1
• Time in Patent Office: 1,023 Days
• CPC Class Codes: G06V 20/54 of traffic, e.g. cars on the road, trains or boats
• CPC Class Codes: G08G 1/0116 from roadside infrastructure, e.g. beacons
• CPC Class Codes: G08G 1/0145 for active traffic flow control
• CPC Class Codes: G08G 1/042 using inductive or magnetic detectors

Abstract

Systems and method are disclosed for adaptive and/or autonomous traffic control. In one illustrative implementation, there is provided a method for processing traffic information. Moreover, the method may include receiving data regarding travel of vehicles associated with an intersection, using neural network technology to recognize types and/or states of traffic, and using the neural network technology to process/determine/memorize optimal traffic flow decisions as a function of experience information. Exemplary implementations may also include using the neural network technology to achieve efficient traffic flow via recognition of the optimal traffic flow decisions.

Claims

1. A method for processing traffic information, the method comprising:

receiving data regarding travel of vehicles associated with an intersection;

using a neural network to recognize traffic types associated with the intersection, the traffic types being composed of one or both of traffic context types and traffic category types;

using the neural network to recognize traffic states associated with the intersection;

using the neural network to process optimal traffic flow decisions for the intersection as a function of prior traffic information regarding the intersection;

using the neural network to achieve efficient traffic flow for the intersection via recognition and use of the optimal traffic flow decisions in providing an output that controls a traffic light system associated with the intersection; and

performing recognition of the traffic types using an array having a plurality of sensor inputs from video sensors, physical sensors in infrastructure of the intersection and/or the traffic light system, and other sensor inputs from one or more electromagnetic sensors;

wherein the other sensor inputs include an infrared traffic input providing capability of detecting vehicle type and passenger quantity from transmission of electromagnetic spectra data types;

wherein the infrared traffic input comprises;

infrared signatures of detected traffic objects independent from adverse weather or light conditions for recognition of vehicle position and vehicle type; and

detection of passenger occupancy of the vehicles which is used in higher level traffic flow decision making;

and wherein the method further comprises;

giving priority to passenger-weighted incoming traffic.

2. The method of claim 1 further comprising:

processing new sensory inputs; and

retraining previously deployed neurons to take advantage of the new sensors;

wherein recognition of the traffic types is enhanced via adapting recognition by the previously deployed neurons to utilize the new inputs.

3. The method of claim 1 further comprising:

processing new sensory inputs; and

utilizing additional, reserve neuron capacity that was initially included in the system, wherein the reserve neuron capacity is subsequently trained to utilize the new inputs.

4. The method of claim 1 further comprising:

processing the other sensor inputs to detect small traffic objects defined as traffic objects that are smaller than cars, wherein the system is configured to adapt to the small traffic objects; and

incorporating the small traffic object information into the higher level traffic flow control decision-making.

5. The method of claim 1 further comprising:

utilizing 2 or more layers of neural network neuron storage elements for unique recognition, classification, and/or traffic flow decision-making tasks.

6. The method of claim 5 wherein the unique recognition includes feeding from a low level layer to higher level layers.

7. The method of claim 5 wherein the traffic flow decision-making includes state information inputs from nearby similar traffic controllers comparable with intermediate level traffic state recognition at the local controller which are combined at higher levels to optimize traffic flow decision-making at the local traffic signal light.

8. A method for processing traffic information, the method comprising:

receiving data regarding travel of vehicles associated with an intersection;

using a neural network to recognize traffic types associated with the intersection, the traffic types being composed of one or both of traffic context types and traffic category types;

using the neural network to recognize traffic states associated with the intersection;

using the neural network to process optimal traffic flow decisions for the intersection as a function of prior traffic information regarding the intersection; and

using the neural network to achieve efficient traffic flow for the intersection via recognition and use of the optimal traffic flow decisions in providing an output that controls a traffic light system associated with the intersection;

wherein higher levels of neural network recognition hierarchy are used to determine relative traffic object and pedestrian positions and speeds, which is used to select optimal traffic light sequences using a sequence of neural network recognition cycles.

9. The method of claim 8 further comprising performing recognition of the traffic types using an array having a plurality of sensor inputs from video sensors, physical sensors in infrastructure of the traffic light system, and other sensor inputs from one or more electromagnetic sensors.

10. The method of claim 9 wherein the other sensor inputs include an infrared traffic input providing detection capability of vehicle type and passenger quantity from transmission of various electromagnetic spectra data types.

11. The method of claim 10 wherein the infrared traffic input comprises:

infrared signatures of detected traffic objects independent from adverse weather or light conditions for recognition of vehicle position and vehicle type; and

detection of vehicle passenger occupancy which is used in higher level traffic flow decision making;

and wherein the method further comprises;

giving priority to passenger-weighted incoming traffic.

12. The method of claim 8 further comprising:

processing new sensory inputs; and

retraining previously deployed neurons to take advantage of the new sensors;

wherein recognition of the traffic types is enhanced via adapting recognition by the previously deployed neurons to utilize the new inputs.

13. The method of claim 8 further comprising:

processing new sensory inputs;

utilizing additional, reserve neuron capacity that was initially included in the system, wherein the reserve neuron capacity is subsequently trained to utilize the new inputs.

14. The method of claim 8 further comprising:

processing one or more electromagnetic sensor inputs to detect reduced traffic flow, wherein the system is configured to adapt to the reduced traffic flow; and

incorporating the reduced traffic flow information into the higher level traffic flow control decision-making.

15. The method of claim 8 further comprising:

utilizing 2 or more layers of neural network neuron storage elements for unique recognition, classification, and/or traffic flow decision making tasks wherein the unique recognition includes feeding from a low-level layer to higher level layers.

16. The method of claim 8 further comprising:

utilizing 2 or more layers of neural network neuron storage elements for one or both of classification and/or traffic flow decision-making tasks.

17. The method of claim 16 further comprising performing recognition of the traffic types using an array having a plurality of sensor inputs from video sensors, and physical sensors in infrastructure of the traffic light system, and other sensor inputs from one or more electromagnetic sensors.

18. The method of claim 17 wherein the other sensor inputs include an infrared traffic input providing capability of detecting vehicle type and passenger quantity from transmission of electromagnetic spectra data types.

19. The method of claim 18 wherein the infrared traffic input comprises:

infrared signatures of detected traffic objects independent from adverse weather or light conditions for recognition of vehicle position and vehicle type; and

detection of passenger occupancy which is used in higher level traffic flow decision making;

and wherein the method further comprises;

giving priority to passenger-weighted incoming traffic.

20. The method of claim 8 further comprising:

utilizing higher levels of neural network recognition hierarchy to prioritize passage through the intersection on the basis of one or more of vehicle priority, passenger capacity, vehicle size, and/or vehicle weight.