Multi-spectral LIDAR object tracking

1. A method for object tracking, wherein each step is performed by a multispectral lidar system, comprising the steps of:

transmitting, with the multispectral lidar system, a first multispectral light beam;

analyzing a response, of a photodetector, to a return of the first multispectral light beam;

based on the analyzing of the response: (i) generating a point cloud with a set of points; and (ii) determining a set of spectral responses of the set of points;

normalizing, based on a distance to the points in the point cloud, the set of spectral responses of the set of points in the point cloud into a set of normalized spectral responses;

determining a set of material compositions for the set of points using the set of normalized spectral responses;

based on both: (i) the set of material compositions; and (ii) a geometric aspect of the set of points: grouping points in the set of points into two groups;

creating two tracking object definitions using the two groups, wherein the two tracking object definitions include associated material composition information from the set of material compositions and geometric information of the geometric aspect;

transmitting, with the multispectral lidar system and after transmitting the first multispectral light beam, a second multispectral light beam;

tracking two physical objects using: (i) a response of the photodetector to the second multispectral light beam; and (ii) the two tracking object definitions;

updating the two tracking object definitions for further tracking using the response of the photodetector to the second multispectral light beam;

wherein the tracking step is different than the grouping step because: (i) the tracking step uses the two tracking object definitions to identify the two physical objects from the response to the second multispectral light beam; and (ii) the grouping step does not use any object definitions to group the set of points into two groups from the response to the first multispectral light beam;

wherein the two tracking object definitions are definitions for the two physical objects;

wherein the set of material compositions includes two different material compositions corresponding to the two physical objects;

wherein the two physical objects are contiguous in the point cloud; and

wherein tracking the two physical objects includes separable tracking of the two physical objects.

2. The method of object tracking from claim 1 , wherein:

the multispectral lidar system is embedded on a vehicle; and

the two physical objects are potential obstacles for the vehicle.

3. The method of object tracking from claim 1 , wherein:

the multispectral lidar system is a surveillance system; and

the two physical objects are moving objects in a surveillance area.

4. The method of object tracking from claim 1 , further comprising:

determining that one of the two physical objects is occluded using: (i) a response of the photodetector; and (ii) the two tracking object definitions.

5. The method of object tracking from claim 1 , wherein:

analyzing the response comprises: applying a set of wavelength filters to the response; and

the set of wavelength filters include a bandpass filter which passes infrared light.

6. The method of object tracking from claim 1 , further comprising:

storing a library of spectral responses; and

wherein determining the set of material compositions for the set of points uses the library of spectral responses.

7. The method of object tracking from claim 6 , wherein:

the set of spectral responses of the set of points is a set of response vectors;

the library of spectral responses is a library of stored vectors; and

determining the set of material compositions for the set of points comprises minimum distance analyses of the set of response vectors and the library of stored vectors.

8. The method of object tracking from claim 1 , further comprising:

detecting an atmospheric condition;

wherein the normalizing of the set of spectral responses uses the atmospheric condition.

9. The method of object tracking from claim 1 , wherein:

the two tracking object definitions are two generic object definitions; and

the steps used to generate the two tracking object definitions do not include any object classifications.

10. The method of object tracking from claim 1 , wherein:

the step of grouping points in the set of points into two groups only uses the set of material compositions and a set of distances; and

the step of tracking two physical objects does not include any classification of the two physical objects.

11. A multispectral lidar system for object tracking, comprising:

a multispectral light beam transmitter that transmits a first multispectral light beam and a second multispectral light beam;

a photodetector arranged to receive a return of the first multispectral light beam and a return of the second multispectral light beam;

a segmentation processing system programmed to:

analyze a response, of the photodetector, to the return of the first multispectral light beam;

based on the analyzing of the response to the return of the first multispectral light beam: (i) generate a point cloud with a set of points; and (ii) determine a set of spectral responses of the set of points;

normalize, based on a distance to the points in the point cloud, the set of spectral responses of the set of points in the point cloud into a set of normalized spectral responses;

based on the set of spectral responses of the set of points: determine a set of material compositions for the set of points, wherein determining the set of material compositions for the set of points uses the set of normalized spectral responses;

based on both: (i) the set of material compositions; and (ii) a geometric aspect of the set of points: group points in the set of points into two groups; and

create two tracking object definitions using the two groups, wherein the tracking object definitions include both material composition information and geometric aspect information;

a tracking processing system programmed to:

analyze the response, of the photodetector, to the return of the second multispectral light beam;

track, based on the analyzing of the response to the return of the second multispectral light beam and using the two tracking object definitions, two physical objects;

update the two tracking object definitions for further tracking using the response of the photodetector to the second multispectral light beam;

wherein the two tracking object definitions are definitions for the two physical objects;

wherein the tracking by the tracking processing system is different than the grouping by the segmentation processing system because: (i) the tracking uses the two tracking object definitions to identify the two physical objects from the response to the second multispectral light beam; and (ii) the grouping does not use any object definitions to group the set of points into two groups from the response to the first multispectral light beam;

wherein the set of material compositions includes two different material compositions corresponding to the two physical objects;

wherein the two physical objects are contiguous in the point cloud; and

wherein tracking the two physical objects includes separable tracking of the two physical objects.

12. The multispectral lidar system for object tracking from claim 11 , wherein:

the multispectral lidar system is embedded on a vehicle; and

the two physical objects are potential obstacles for the vehicle.

13. The multispectral lidar system for object tracking from claim 11 , wherein:

the multispectral lidar system is a surveillance system; and

the two physical objects are moving objects in a surveillance area.

14. The multispectral lidar system for object tracking from claim 11 , wherein the tracking processing system is further programmed to:

determine that one of the two physical objects is occluded using: (i) a response of the photodetector; and (ii) the two tracking object definitions.

15. The multispectral lidar system for object tracking from claim 11 , wherein:

analyzing the responses comprises: applying a set of wavelength filters to the response; and

the set of wavelength filters include a bandpass filter which passes infrared light.

16. The multispectral lidar system for object tracking from claim 15 , further comprising:

a memory that stores a library of spectral responses; and

wherein determining the set of material compositions for the set of points uses the library of spectral responses.

17. The multispectral lidar system for object tracking from claim 11 , further comprising:

a memory that stores a library of spectral responses; and

wherein determining the set of material compositions for the set of points uses the library of spectral responses.

18. The multispectral lidar system for object tracking from claim 17 , wherein:

the set of spectral responses is a set of response vectors;

the library of spectral responses is a library of stored vectors; and

the determining of the set of material compositions for the set of points comprises minimum distance analyses of the set of response vectors and the library of stored vectors.

19. The multispectral lidar system for object tracking from claim 11 , further comprising:

a memory that stores a library of spectral responses; and

wherein determining the set of material compositions for the set of points uses the library of spectral responses.

20. The multispectral lidar system for object tracking from claim 11 , wherein the segmentation processing system is further programmed to:

detect an atmospheric condition;

wherein the normalizing of the set of spectral responses uses the atmospheric condition.

21. The multispectral lidar system for object tracking from claim 11 , wherein:

the two tracking object definitions are two generic object definitions; and

the segmentation processing system does not conduct any classifications of the tracked objects.

22. The multispectral lidar system for object tracking from claim 11 , wherein:

the grouping of points in the set of points into two groups only uses the set of material compositions and the geometric aspect; and

the tracking processing system does not conduct any classification of the two physical objects.