Method and system for detecting lane line based on lidar data

1 . A method of detecting a lane line based on lidar data, the method being executed by a computing device mounted on a vehicle and coupled to a lidar sensor, the computing device comprising a processor, the method comprising:

scanning, by a lidar sensor mounted on a vehicle, a surrounding environment of the vehicle, and generating, by the lidar sensor, lidar data;

detecting, by a processor mounted on the vehicle and coupled to the lidar sensor through an internal communication bus, points each estimated as a lane line in the lidar data;

dividing, by the processor, a region including the detected points into grids at regular intervals, and averaging positions of the detected points included in each grid to generate average points;

performing, by the processor, a sampling operation of randomly sampling points from the average points;

identifying, by the processor, symmetry of the detected points based on an arbitrary line;

performing, by the processor, an estimation operation of estimating parameters of a mathematical model using the sampled points;

performing, by the processor, a setting operation of calculating distances between each of the detected points and the mathematical model in which the parameters are estimated and setting the calculated distances as scores;

when the symmetry of the detected points is not identified, assigning, by the processor, a weight to points detected in front of the lidar sensor and generating the scores based on the distances and the weight;

performing, by the processor, a summation operation of summing the scores; and

setting, by the processor, the mathematical model determined according to the summation score as a lane line,

wherein the sampling operation, the estimation operation, the setting operation, and the summation operation are repeatedly performed a predetermined number of times and parameters of the mathematical model having a highest summation score are selected.

2 . The method of claim 1 , wherein the processor assigns a higher score as the distances between each of the detected points and the mathematical model in which the parameters are estimated are shorter.

3 . The method of claim 1 , wherein the identifying of, by the processor, the symmetry of the detected points based on the arbitrary line with respect to the detected points includes:

classifying, by the processor, the detected points into first points positioned above the arbitrary line and second points positioned below the arbitrary line based on the arbitrary line;

overlapping, by the processor, the first points and the second points and determining whether thicknesses of the overlapping first points and second points are greater than or equal to a predetermined length; and

when it is determined that the thicknesses of the overlapping first points and second points are greater than or equal to the predetermined length, determining that, by the processor, the detected points have no symmetry.

4 . A system for detecting a lane line based on lidar data, the system comprising:

a lidar sensor mounted on a vehicle and configured to scan a surrounding environment of the vehicle and generate lidar data; and

a computing device mounted on the vehicle and coupled to the lidar sensor through an internal communication bus,

wherein the computing device includes:

a processor; and

a memory in which instructions executed by the processor are stored, and

the instructions are implemented to:

detect points each estimated as a lane line in a lidar data,

divide a region including the detected points into grids at regular intervals, and average positions of points included in each grid to generate average points to compensate for point density variations based on a distance from a lidar sensor,

perform a sampling operation of randomly sampling points from the average points to generate sampled points,

identify symmetry of the detected points based on an arbitrary line,

perform an estimation operation of estimating parameters of a mathematical model using the detected points,

perform a setting operation of calculating distances between each of the detected points and the mathematical model in which the parameters are estimated and setting the calculated distances as scores,

when the symmetry is not identified, assign a weight to points detected in front of the lidar sensor and generate the scores based on the distances and the weight,

perform a summation operation of summing the scores, and

set the mathematical model determined according to the summation score as a lane line,

repeatedly perform the sampling operation, the estimation operation, the setting operation, and the summation operation a predetermined number of times, and

select parameters of the mathematical model having a highest summation score.

5 . The system of claim 4 , wherein the instructions are implemented to assign a higher score as the distances between each of the detected points and the mathematical model in which the parameters are estimated are shorter.

6 . The system of claim 4 , further comprising:

instructions implemented to identify symmetry of the detected points based on an arbitrary line with respect to the detected points; and

when the symmetry of the detected points is not identified, instructions implemented to assign a weight to the points detected in front of the lidar sensor, calculate distances between each of the detected points and the mathematical model in which the parameters are estimated, multiply the points by the calculated distances, and set results obtained by multiplication as the scores.

7 . The system of claim 4 , wherein the instructions that identify the symmetry of the detected points based on the arbitrary line with respect to the detected points are implemented to:

classify the detected points into first points positioned above the arbitrary line and second points positioned below the arbitrary line based on the arbitrary line;

overlap the first points and the second points and determine whether thicknesses of the first points and the second points are greater than or equal to a predetermined length; and

when it is determined that the thicknesses of the overlapping first points and second points are greater than or equal to the predetermined length, determine that the detected points have no symmetry.