Real-time localization error correction of autonomous vehicle

1. A processor for real-time localization error correction of an autonomous vehicle (AV), configured to:

retrieve a reference landmark around the AV from a map aggregating server (MAS), wherein the AV is configured to interact with the MAS for real-time localization;

detect, in real time, a ground truth landmark corresponding to the reference landmark, according to image data captured by one or more image capture devices installed on the AV;

calculate a deviation between the ground truth landmark and the reference landmark as a real-time correction value for the real-time localization of the AV;

retrieve, from the MAS, a reference correction value associated with a road segment where the AV is travelling;

calculate an average value of the reference correction value and the real-time correction value as a final real-time correction value for the real-time localization;

perform the real-time localization error correction of the AV based on the final real-time correction value; and

cause the AV to autonomously drive to a destination based on the real-time localization of the AV after the real-time localization error correction.

2. The processor according to claim 1 , further configured to:

retrieve a number N of reference landmarks around the AV from the MAS;

detect, in real time, a number N of ground truth landmarks respectively corresponding to the number N of reference landmarks;

compare each pair of reference landmark and corresponding ground truth landmark to obtain a number N of deviations between the number N of reference landmarks and the number N of corresponding ground truth landmarks; and

determine an average value of the number N of deviations as the real-time correction value for the real-time localization of the AV, where N is an integer greater than 1.

3. The processor according to claim 1 , further configured to:

report the real-time correction value to the MAS via a communication interface between the MAS and the AV.

4. The processor according to claim 1 , wherein the reference correction value is a real-time correction value reported to the MAS by another AV passing the road segment.

5. The processor according to claim 1 , wherein the reference correction value is an average value of a number M of real-time correction values reported respectively by a number M of AVs passing the road segment during a predetermined period of time, where M is an integer greater than 1.

6. The processor according to claim 1 , further configured to:

report the final real-time correction value to the MAS via a communication interface between the MAS and the AV.

7. The processor according to claim 1 , wherein the reference landmark comprises a road edge or a lane boundary stored and updated periodically in the MAS, and the ground truth landmark comprises a corresponding road edge or a corresponding lane boundary detected by the AV in real time according to the image data captured by the one or more image capture devices installed on the AV.

8. An autonomous vehicle (AV), comprising a processor for real-time localization error correction of the AV, and one or more image capture devices, wherein

the processor is configured to:

retrieve a reference landmark around the AV from a map aggregating server (MAS), wherein the AV is configured to interact with the MAS for real-time localization;

detect, in real time, a ground truth landmark corresponding to the reference landmark, according to image data captured by the one or more image capture devices;

calculate a deviation between the ground truth landmark and the reference landmark as a real-time correction value for the real-time localization of the AV;

retrieve, from the MAS, a reference correction value associated with a road segment where the AV is travelling;

calculate an average value of the reference correction value and the real-time correction value as a final real-time correction value for the real-time localization;

perform the real-time localization error correction of the AV based on the final real-time correction value; and

cause the AV to autonomously drive to a destination based on the real-time localization of the AV after the real-time localization error correction.

9. A method for real-time localization error correction of an autonomous vehicle (AV), comprising:

retrieving a reference landmark around the AV from a map aggregating server (MAS), wherein the AV is configured to interact with the MAS for real-time localization;

detecting, in real time, a ground truth landmark corresponding to the reference landmark, according to image data captured by one or more image capture devices installed on the AV;

calculating a deviation between the ground truth landmark and the reference landmark as a real-time correction value for the real-time localization of the AV;

retrieving, from the MAS, a reference correction value associated with a road segment where the AV is travelling;

calculating an average value of the reference correction value and the real-time correction value as a final real-time correction value for the real-time localization;

performing the real-time localization error correction of the AV based on the final real-time correction value; and

causing the AV to autonomously drive to a destination based on the real-time localization of the AV after the real-time localization error correction.

10. The method according to claim 9 , further comprising:

retrieving a number N of reference landmarks around the AV from the MAS;

detecting, in real time, a number N of ground truth landmarks respectively corresponding to the number N of reference landmarks;

comparing each pair of reference landmark and corresponding ground truth landmark to obtain a number N of deviations between the number N of reference landmarks and the number N of corresponding ground truth landmarks; and

determining an average value of the number N of deviations as the real-time correction value for the real-time localization of the AV, where N is an integer greater than 1.

11. The method according to claim 9 , further comprising:

reporting the real-time correction value to the MAS via a communication interface between the MAS and the AV.

12. The method according to claim 9 , wherein the reference correction value is a real-time correction value reported to the MAS by another AV passing the road segment.

13. The method according to claim 9 , wherein the reference correction value is an average value of a number M of real-time correction values reported respectively by a number M of AVs passing the road segment during a predetermined period of time, where M is an integer greater than 1.

14. The method according to claim 9 , further comprising:

reporting the final real-time correction value to the MAS via a communication interface between the MAS and the AV.

15. The method according to claim 9 , wherein the reference landmark comprises a road edge or a lane boundary stored and updated periodically in the MAS, and the ground truth landmark comprises a corresponding road edge or a corresponding lane boundary detected by the AV in real time according to the image data captured by the one or more image capture devices installed on the AV.

16. A non-transitory computer-readable medium having instructions stored thereon, wherein the instructions, when executed by a processor, cause the processor to perform the method for the real-time localization error correction of the AV according to claim 9 .