Method and a server for updating a map representation

1. A method of updating a candidate portion of a map representation of a geographical region, the map representation having been built based on point cloud data captured at least in part by a Light Detection and Ranging (LIDAR) system of a Self-Driving Car (SDC) that has operated in the geographical region, the method being executable by a server, the method comprising:

receiving an initial approximation location of the SDC in the candidate portion, the initial approximation location being associated with a timestamp;

generating, by the server using a localization algorithm, three or more candidate locations in the candidate portion, each candidate location is based on a respective point cloud, associated with the candidate portion, and the initial approximation location;

determining, by the server using a convergence metric, a first circle for evaluating quality of localization of the localization algorithm, the first circle having a first radius,

determining a second circle having a second radius, the second circle is a smallest circle encompassing the three or more candidate locations;

identifying, by the server, the candidate portion of the map representation as an outdated portion of the map representation, the identifying being done in response to the second radius being greater than the first radius;

updating, by the server, the outdated portion of the map representation using the respective point cloud of the each candidate location; and

controlling, based on the first circle, operation of the SDC.

2. The method of claim 1 , wherein the first circle encompassing the three or more candidate locations is an indicative of converging of the three or more candidate locations.

3. The method of claim 2 , wherein a first candidate location of the three or more candidate locations is a first candidate position of the SDC.

4. The method of claim 3 , wherein the each candidate location being associated with a respective candidate orientation of the SDC, and wherein the candidate portion of the map representation is identified as the outdated portion of the map representation if at least one of:

(i) the three or more candidate positions are not within the first circle, and

(ii) the respective candidate orientation of the each candidate location is not within a sector of a pre-determined angle.

5. A method of updating a candidate portion of a map representation of a geographical region, the map representation having been built based on point cloud data captured at least in part by a first Light Detection and Ranging (LIDAR) system of a first Self-Driving Car (SDC) that has operated in the geographical region, the method being executable by a server, the method comprising:

receiving an initial approximation location of the first SDC in the candidate portion, the initial approximation location being associated with a timestamp;

determining, by the server, a first point cloud from a plurality of point clouds that has been captured by the LIDAR system at the timestamp;

generating, by the server using a localization algorithm, a first candidate location in the candidate portion, the first candidate location is based on the initial approximation location, the first point cloud, and the map representation;

determining, by the server, a second point cloud associated with the candidate portion;

generating, by the server using the localization algorithm, a second candidate location in the candidate portion, the second candidate location is based on the initial approximation location, the second point cloud, and the map representation;

determining, by the server, a third point cloud associated with the candidate portion;

generating, by the server using the localization algorithm, a third candidate location in the candidate portion, the third candidate location is based at least in part on the initial approximation location, the third point cloud, and the map representation;

determining, by the server using a convergence metric, a first circle for evaluating quality of localization of the localization algorithm in the candidate portion of the map representation, the first circle having a first radius;

determining a second circle having a second radius, the second circle is a smallest circle encompassing the first, the second, and the third candidate location;

identifying, by the server, the candidate portion of the map representation as an outdated portion of the map representation, the identifying being done in response to the second radius being greater than the first radius;

updating, by the server, the outdated portion of the map representation using the first, the second, and the third point cloud; and

controlling, based on the first circle, operation of the first SDC.

6. The method of claim 5 , wherein the determining the second point cloud comprises generating, by the server, a simulated point cloud using the first point cloud, the simulated point cloud being a degraded version of the first point cloud.

7. The method of claim 6 , wherein the degraded version of the first point cloud is missing a subset of points from the first point cloud.

8. The method of claim 6 , wherein the generating the simulated point cloud comprises applying, by the server, a filter representing an effect of an environmental condition on the first point cloud.

9. The method of claim 8 , wherein the environmental condition is at least one of rain, snow, and dirt occluding the LIDAR system.

10. The method of claim 5 , wherein the determining the second point cloud comprises determining, by the server, the second point cloud from log data of a second SDC that has operated in the geographical region, the second SDC having a second LIDAR system, the second point cloud having been captured by the second LIDAR system when the the second SDC has been located at the candidate portion.

11. The method of claim 5 , wherein the convergence metric includes a trained Machine Learning Algorithm (MLA).

12. The method of claim 11 further comprising:

prior said determining the first circle, generating, by the server a training set including a training input and a training label, the training input including the map representation, the first candidate location, and the first point cloud; and

training, by the server, an untrained MLA to output the trained MLA.

13. The method of claim 11 further comprises:

prior said determining the first circle, generating, by the server a training set including a training input and a training label, the training input including the map representation, the second candidate location, and the second point cloud; and

training, by the server, an untrained MLA to output the trained MLA.

14. The method of claim 5 , wherein the first, the second, and the third candidate location being associated with respective candidate orientations of the SDC, and wherein the method further comprising

determining, by the server using a second convergence metric, a sector of a predetermined angle for evaluating quality of localization of the localization algorithm in the candidate portion of the map representation; and

identifying, by the server, the candidate portion of the map representation as an outdated portion of the map representation, the identifying being done in response to one or more of the respective candidate orientations being not within the sector.

15. A server for updating a candidate portion of a map representation of a geographical region, the map representation having been built based on point cloud data captured at least in part by a Light Detection and Ranging (LIDAR) system of a Self-Driving Car (SDC) that has operated in the geographical region, the server being configured to:

receive an initial approximation location of the SDC in the candidate portion, the initial approximation location being associated with a timestamp;

determine a first point cloud from a plurality of point clouds that has been captured by the LIDAR system at the timestamp;

generate, using a localization algorithm, a first candidate location in the candidate portion, the first candidate location is based on the initial approximation location, the first point cloud, and the map representation;

determine a second point cloud associated with the candidate portion;

generate, using the localization algorithm, a second candidate location in the candidate portion, the second candidate location is based on the initial approximation location, the second point cloud, and the map representation;

determine a third point cloud associated with the candidate portion;

generate, using the localization algorithm, a third candidate location in the candidate portion, the third candidate location is based on the initial approximation location, the third point cloud, and the map representation;

determine, using a convergence metric, a first circle for evaluating quality of localization of the localization algorithm in the candidate portion of the map representation, the first circle having a first radius;

determine a second circle having a second radius, the second circle is a smallest circle encompassing the first, the second, and the third candidate location;

identify, the candidate portion of the map representation as an outdated portion of the map representation, said identification being done in response to the second radius being greater than the first radius;

update the outdated portion of the map representation using the first, the second, and the third point cloud; and

control, based on the first circle, operation of the SDC.

16. The server of claim 15 , wherein to determine the second point cloud comprises the server configured to generate a simulated point cloud using the first point cloud, the simulated point cloud being a degraded version of the first point cloud.

17. The server of claim 16 , wherein the degraded version of the first point cloud is missing a subset of points from the first point cloud.

18. The server of claim 16 , wherein to generate the simulated point cloud comprises the server configured to apply a filter representing an effect of an environmental condition on the first point cloud.

19. The server of claim 18 , wherein the environmental condition is at least one of rain, snow, and dirt occluding the LIDAR system.