Systems and methods for controlling the operation of an autonomous vehicle using multiple traffic light detectors

1. A system for controlling operation of an autonomous vehicle, the system comprising:

one or more processors; and

a memory communicably coupled to the one or more processors and storing:

a sensor-based traffic light detection module including instructions that when executed by the one or more processors cause the one or more processors to perform sensor-based traffic light detection at an intersection to produce a sensor-based detection output, the sensor-based detection output having an associated first confidence level;

a vehicle-to-infrastructure-based (V2I-based) traffic light detection module including instructions that when executed by the one or more processors cause the one or more processors to perform V2I-based traffic light detection at the intersection to produce a V2I-based detection output, the V2I-based detection output having an associated second confidence level;

a fusion module including instructions that when executed by the one or more processors cause the one or more processors to select as a final traffic-light-detection output whichever of the sensor-based detection output and the V2I-based detection output has a higher associated confidence level subject to an override based on consideration of a risk associated with a transition through the intersection that the autonomous vehicle plans to execute; and

a control module including instructions that when executed by the one or more processors cause the one or more processors to control the operation of the autonomous vehicle based, at least in part, on the final traffic-light-detection output.

2. The system of claim 1 , wherein the sensor-based traffic light detection module includes instructions to produce the sensor-based detection output by analyzing image data associated with the intersection.

3. The system of claim 1 , wherein the V2I-based traffic light detection module includes instructions to compute the associated second confidence level, at least in part, by comparing V2I signals received from an information system of the intersection with environmental sensor data associated with the intersection using a learning-based classifier.

4. The system of claim 1 , wherein the V2I-based traffic light detection module includes instructions to compute the associated second confidence level, at least in part, by processing past sensor-based detection output data and past V2I-based detection output data using a learning-based classifier.

5. The system of claim 1 , wherein the instructions in the control module to control the operation of the autonomous vehicle based, at least in part, on the final traffic-light-detection output include instructions to control one or more of steering, acceleration, and braking.

6. The system of claim 1 , wherein the final traffic-light-detection output includes one or more transitions, the one or more transitions corresponding to different possible paths through the intersection, and an estimated traffic light state for each of the one or more transitions.

7. The system of claim 6 , wherein the final traffic-light-detection output further includes state-timing information for at least one traffic light at the intersection and an overall confidence level for the final traffic-light-detection output.

8. A non-transitory computer-readable medium for controlling operation of an autonomous vehicle and storing instructions that when executed by one or more processors cause the one or more processors to:

perform sensor-based traffic light detection at an intersection to produce a sensor-based detection output, the sensor-based detection output having an associated first confidence level;

perform vehicle-to-infrastructure-based (V2I-based) traffic light detection at the intersection to produce a V2I-based detection output, the V2I-based detection output having an associated second confidence level;

select as a final traffic-light-detection output whichever of the sensor-based detection output and the V2I-based detection output has a higher associated confidence level subject to an override based on consideration of a risk associated with a transition through the intersection that the autonomous vehicle plans to execute; and

control the operation of the autonomous vehicle based, at least in part, on the final traffic-light-detection output.

9. The non-transitory computer-readable medium of claim 8 , wherein the instructions include instructions to compute the associated second confidence level, at least in part, by comparing V2I signals received from an information system of the intersection with environmental sensor data associated with the intersection using a learning-based classifier.

10. The non-transitory computer-readable medium of claim 8 , wherein the instructions include instructions to compute the associated second confidence level, at least in part, by processing past sensor-based detection output data and past V2I-based detection output data using a learning-based classifier.

11. The non-transitory computer-readable medium of claim 8 , wherein the final traffic-light-detection output includes one or more transitions, the one or more transitions corresponding to different possible paths through the intersection, and an estimated traffic light state for each of the one or more transitions.

12. The non-transitory computer-readable medium of claim 11 , wherein the final traffic-light-detection output further includes state-timing information for at least one traffic light at the intersection and an overall confidence level for the final traffic-light-detection output.

13. A method of controlling operation of an autonomous vehicle, the method comprising:

performing traffic light detection at an intersection using a sensor-based traffic light detector to produce a sensor-based detection output, the sensor-based detection output having an associated first confidence level;

performing traffic light detection at the intersection using a vehicle-to-infrastructure-based (V2I-based) traffic light detector to produce a V2I-based detection output, the V2I-based detection output having an associated second confidence level;

selecting as a final traffic-light-detection output whichever of the sensor-based detection output and the V2I-based detection output has a higher associated confidence level subject to an override based on consideration of a risk associated with a transition through the intersection that the autonomous vehicle plans to execute; and

controlling the operation of the autonomous vehicle based, at least in part, on the final traffic-light-detection output.

14. The method of claim 13 , wherein the associated second confidence level is computed, at least in part, by comparing V2I signals received from an information system of the intersection with environmental sensor data associated with the intersection using a learning-based classifier.

15. The method of claim 13 , wherein the associated second confidence level is computed, at least in part, by processing past sensor-based detection output data and past V2I-based detection output data using a learning-based classifier.

16. The method of claim 13 , wherein the final traffic-light-detection output includes one or more transitions, the one or more transitions corresponding to different possible paths through the intersection, and an estimated traffic light state for each of the one or more transitions.

17. The method of claim 16 , wherein the final traffic-light-detection output further includes timing information for at least one traffic light at the intersection and an overall confidence level for the final traffic-light-detection output.