Method and apparatus for identifying static obstacle

1. A method for identifying a static obstacle, comprising:

generating, in response to finding out a historical laser point cloud sequence of an obstacle, a detected laser point cloud of the obstacle in a current laser point cloud frame belonging to the historical laser point cloud sequence of the obstacle, determining information corresponding to the historical laser point cloud sequence of the obstacle, wherein the detected laser point cloud of the obstacle and a historical laser point cloud of the obstacle in the historical laser point cloud sequence of the obstacle represent a given obstacle, and the determining information comprises: a similarity between a historical motion characteristic of the given obstacle and a noise type motion characteristic, a matching degree between an appearance characteristic of the detected laser point cloud of the obstacle and an appearance characteristic of the historical laser point cloud of the obstacle, and an overlap ratio between the detected laser point cloud of the obstacle and the historical laser point cloud of the obstacle; and

determining whether the given obstacle is a static obstacle based on the determining information.

2. The method according to claim 1 , wherein the historical motion characteristic comprises: a historical speed distribution of the given obstacle obtained through a motion estimation, a historical location distribution of the given obstacle, a historical speed variation of the given obstacle obtained through a motion estimation within a preset time length, and a historical location variation of the given obstacle within the preset time length.

3. The method according to claim 2 , wherein the noise type motion characteristic comprises: a noise type distribution obtained based on a statistical rule, and a noise data variation within the preset time length obtained based on a kinematic rule.

4. The method according to claim 3 , wherein the appearance characteristic comprises: a size, a laser point density, and a geometrical shape.

5. The method according to claim 4 , wherein the determining whether the given obstacle is a static obstacle based on the determining information comprises:

determining a displacement amount corresponding to the historical laser point cloud of the obstacle based on a historical speed of the given obstacle obtained through a motion estimation at a moment of collecting a historical laser point cloud frame, the historical laser point cloud of the obstacle belonging to the historical laser point cloud frame, and a length of the collection period using a lidar;

respectively calculating an Intersection-over-Union between the detected laser point cloud of the obstacle and the historical laser point cloud of the obstacle and an Intersection-over-Union between the detected laser point cloud of the obstacle and the historical laser point cloud of the obstacle after moving the displacement amount; and

calculating an overlap ratio between the detected laser point cloud of the obstacle and the historical laser point cloud of the obstacle based on the Intersection-over-Union between the detected laser point cloud of the obstacle and the historical laser point cloud of the obstacle and the Intersection-over-Union between the detected laser point cloud of the obstacle and the historical laser point cloud of the obstacle after moving the displacement amount.

6. An apparatus for identifying a static obstacle, comprising:

at least one processor; and

a memory storing instructions, the instructions when executed by the at least one processor, cause the at least one processor to perform operations, the operations comprising:

generating, in response to finding out a historical laser point cloud sequence of an obstacle, a detected laser point cloud of the obstacle in a current laser point cloud frame belonging to the historical laser point cloud sequence of the obstacle, determining information corresponding to the historical laser point cloud sequence of the obstacle, wherein the detected laser point cloud of the obstacle and a historical laser point cloud of the obstacle in the historical laser point cloud sequence of the obstacle represent a given obstacle, and the determining information comprises: a similarity between a historical motion characteristic of the given obstacle and a noise type motion characteristic, a matching degree between an appearance characteristic of the detected laser point cloud of the obstacle and an appearance characteristic of the historical laser point cloud of the obstacle, and an overlap ratio between the detected laser point cloud of the obstacle and the historical laser point cloud of the obstacle; and

determining whether the given obstacle is a static obstacle based on the determining information.

7. The apparatus according to claim 6 , wherein the historical motion characteristic comprises: a historical speed distribution of the given obstacle obtained through a motion estimation, a historical location distribution of the given obstacle, a historical speed variation of the given obstacle obtained through a motion estimation within a preset time length, and a historical location variation of the given obstacle within the preset time length.

8. The method according to claim 7 , wherein the noise type motion characteristic comprises: a noise type distribution obtained based on a statistical rule, and a noise data variation within the preset time length obtained based on a kinematic rule.

9. The apparatus according to claim 8 , wherein the appearance characteristic comprises: a size, a laser point density, and a geometrical shape.

10. The apparatus according to claim 9 , wherein the determining whether the given obstacle is a static obstacle based on the determining information comprises:

determining a displacement amount corresponding to the historical laser point cloud of the obstacle based on a historical speed of the given obstacle obtained through a motion estimation at a moment of collecting a historical laser point cloud frame, the historical laser point cloud of the obstacle belonging to the historical laser point cloud frame, and a length of the collection period using a lidar; respectively calculating an Intersection-over-Union between the detected laser point cloud of the obstacle and the historical laser point cloud of the obstacle and an Intersection-over-Union between the detected laser point cloud of the obstacle and the historical laser point cloud of the obstacle after moving the displacement amount; and calculating an overlap ratio between the detected laser point cloud of the obstacle and the historical laser point cloud of the obstacle based on the Intersection-over-Union between the detected laser point cloud of the obstacle and the historical laser point cloud of the obstacle and the Intersection-over-Union between the detected laser point cloud of the obstacle and the historical laser point cloud of the obstacle after moving the displacement amount.

11. A non-transitory computer-readable storage medium storing a computer program, the computer program when executed by one or more processors, causes the one or more processors to perform operations, the operations comprising:

generating, in response to finding out a historical laser point cloud sequence of an obstacle, a detected laser point cloud of the obstacle in a current laser point cloud frame belonging to the historical laser point cloud sequence of the obstacle, determining information corresponding to the historical laser point cloud sequence of the obstacle, wherein the detected laser point cloud of the obstacle and a historical laser point cloud of the obstacle in the historical laser point cloud sequence of the obstacle represent a given obstacle, and the determining information comprises: a similarity between a historical motion characteristic of the given obstacle and a noise type motion characteristic, a matching degree between an appearance characteristic of the detected laser point cloud of the obstacle and an appearance characteristic of the historical laser point cloud of the obstacle, and an overlap ratio between the detected laser point cloud of the obstacle and the historical laser point cloud of the obstacle; and

determining whether the given obstacle is a static obstacle based on the determining information.