Granted Patent B2

US 12,638,554 · App. 18/004,414 · Granted May 26, 2026

Inventors: Qi Kong (Beijing, CN); Xinyu Xu (Beijing, CN)

Assignee: BEIJING JINGDONG QIANSHI TECHNOLOGY CO., LTD.

G01S7/4808G01S17/89G01S17/931

Abstract

Claims (59)

1 . A method for identifying a data point, comprising:

acquiring real-time point cloud data of a current position of a vehicle by a lidar scanning device on the vehicle;

converting a coordinate of a data point in the real-time point cloud data into a first coordinate under a coordinate system of a preset reference point cloud map, wherein the reference point cloud map comprises static data points;

matching the first coordinate of the data point in the real-time point cloud data with the reference point cloud map, comprising:

building a KD-tree according to a coordinate of each data point in the reference point cloud map;

searching in the KD-tree to determine whether there exists a second coordinate identical to the first coordinate of the data point in the real-time point cloud data;

determining the first coordinate of the data point in the real-time point cloud data matches the reference point cloud map under a condition that there exists the second coordinate;

determining whether there exists a third coordinate identical to an adjacent coordinate of the first coordinate of the data point in the real-time point cloud data under a condition that there does not exist the second coordinate;

determining the first coordinate of the data point in the real-time point cloud data matches the reference point cloud map under a condition that there exists the third coordinate; and

determining the data point in the real-time point cloud data as a dynamic data point under a condition that there does not exist the third coordinate; and

determining the data point in the real-time point cloud data as a static data point under a condition that the first coordinate of the data point in the real-time point cloud data matches the reference point cloud map, to perform path planning for the vehicle according to the static data point and the dynamic data point.

2 . The method for identifying the data point according to claim 1 , further comprising:

determining the data point in the real-time point cloud data as a dynamic data point under a condition that the first coordinate of the data point in the real-time point cloud data does not match the reference point cloud map.

3 . The method for identifying the data point according to claim 1 , further comprising:

determining whether the reference point cloud map needs to be updated after acquiring the real-time point cloud data; and

updating the reference point cloud map under a condition that the reference point cloud map needs to be updated.

4 . The method for identifying the data point according to claim 3 , wherein the determining whether the reference point cloud map needs to be updated comprises:

determining whether a distance between the current position of the vehicle and a last position of the vehicle in the reference point cloud map is greater than a preset threshold value; and

determining the reference point cloud map needs to be updated under a condition that the distance between the current position and the last position of the vehicle in the reference point cloud map is greater than the preset threshold value.

5 . The method for identifying the data point according to claim 1 , wherein the matching the first coordinate of the data point in the real-time point cloud data with the reference point cloud map comprises:

building a grid map according to a coordinate of each data point in the reference point cloud map, wherein a grid comprising static data points in the grid map is marked as a static grid, and a grid not comprising the static data points is marked as a dynamic grid;

determining an index of the data point in the real-time point cloud data in the grid map according to the first coordinate of the data point in the real-time point cloud data;

determining whether a first grid determined by the index of the data point in the real-time point cloud data is a static grid; and

determining the first coordinate of the data point in the real-time point cloud data matches the reference point cloud map under a condition that the first grid is a static grid.

6 . The method for identifying the data point according to claim 5 , wherein the matching the first coordinate of the data point in the real-time point cloud data with the reference point cloud map comprises:

taking all neighboring grids of the first grid as second grids under a condition that the first grid is not the static grid;

determining whether the second grids comprises a static grid; and

determining the first coordinate of the data point in the real-time point cloud data matches the reference point cloud map under a condition that the second grids comprises the static grid.

7 . The method for identifying the data point according to claim 6 , wherein the matching the first coordinate of the data point in the real-time point cloud data with the reference point cloud map comprises:

determining the first coordinates of the data point in the real-time point cloud data does not match the reference point cloud map under a condition that the second grids do not comprise the static grid.

8 . The method for identifying the data point according to claim 1 , further comprising:

filtering data points in the real-time point cloud data larger than a preset range.

9 . The method for identifying the data point according to claim 1 , further comprising:

acquiring point cloud map data within a preset range centered on the current position, and

updating the reference point cloud map according to the acquired point cloud map data.

10 . An apparatus for identifying a data point, comprising:

a processor; and

a memory coupled to the processor, storing program instructions which, when executed by the processor, cause the processor to:

acquire real-time point cloud data of a current position of a vehicle by a lidar scanning device on the vehicle;

convert a coordinate of a data point in the real-time point cloud data into a first coordinate under a coordinate system of a preset reference point cloud map, wherein the reference point cloud map comprises static data points;

match the first coordinate of the data point in the real-time point cloud data with the reference point cloud map, comprising:

build a KD-tree according to a coordinate of each data point in the reference point cloud map;

search in the KD-tree to determine whether there exists a second coordinate identical to the first coordinate of the data point in the real-time point cloud data;

determine the first coordinate of the data point in the real-time point cloud data matches the reference point cloud map under a condition that there exists the second coordinate;

determine whether there exists a third coordinate identical to an adjacent coordinate of the first coordinate of the data point in the real-time point cloud data under a condition that there does not exist the second coordinate;

determine the first coordinate of the data point in the real-time point cloud data matches the reference point cloud map under a condition that there exists the third coordinate; and

determine the data point in the real-time point cloud data as a dynamic data point under a condition that there does not exist the third coordinate; and

determine the data point in the real-time point cloud data as a static data point under a condition that the first coordinate of the data point in the real-time point cloud data matches the reference point cloud map, to perform path planning for the vehicle according to the static data point and the dynamic data point.

11 . A non-transitory computer-readable storage medium, which stores a computer program that, when executed by a processor, cause the processor to:

acquire real-time point cloud data of a current position of a vehicle by a lidar scanning device on the vehicle;

convert a coordinate of a data point in the real-time point cloud data into a first coordinate under a coordinate system of a preset reference point cloud map, wherein the reference point cloud map comprises static data points;

match the first coordinate of the data point in the real-time point cloud data with the reference point cloud map, comprising:

build a KD-tree according to a coordinate of each data point in the reference point cloud map;

search in the KD-tree to determine whether there exists a second coordinate identical to the first coordinate of the data point in the real-time point cloud data;

determine the first coordinate of the data point in the real-time point cloud data matches the reference point cloud map under a condition that there exists the second coordinate;

determine whether there exists a third coordinate identical to an adjacent coordinate of the first coordinate of the data point in the real-time point cloud data under a condition that there does not exist the second coordinate;

determine the first coordinate of the data point in the real-time point cloud data matches the reference point cloud map under a condition that there exists the third coordinate; and

determine the data point in the real-time point cloud data as a dynamic data point under a condition that there does not exist the third coordinate; and

determine the data point in the real-time point cloud data as a static data point under a condition that the first coordinate of the data point in the real-time point cloud data matches the reference point cloud map, to perform path planning for the vehicle according to the static data point and the dynamic data point.

Assignments (1)

Priority Claims (1)

Continuity (1)

References Cited (24)