LIDAR systems and methods for detection and classification of objects
A system includes at least one processor configured to detect, based on point cloud information, portions of a particular object, and determine, based on the detected portions, at least a first portion having a first reflectivity corresponding to a license plate, and at least two additional spaced-apart portions corresponding to locations on the particular object other than a location of the first portion. The at least two additional portions have reflectivity substantially lower than the first reflectivity. The at least one processor is further configured to classify the particular object as a vehicle, based on a spatial relationship and a reflectivity relationship between the first portion and the at least two additional portions.
1. A system, the system comprising:
at least one processor configured to:
detect, based on point cloud information, portions of a particular object;
determine, based on the detected portions, at least a first portion having a first reflectivity corresponding to a license plate, and at least two additional spaced-apart portions corresponding to locations on the particular object other than a location of the first portion, and wherein the at least two additional portions have reflectivity substantially lower than the first reflectivity; and
classify the particular object as a vehicle, based on a spatial relationship and a reflectivity relationship between the first portion and the at least two additional portions.
2. The system of claim 1 , wherein the at least one processor is further configured to determine a distance to the particular object and to classify the particular object based on the determined distance and the spatial relationship between the first portion and the at least two additional portions.
3. The system of claim 1 , wherein the at least one processor is further configured to determine an angle of at least one surface associated with the particular object and to classify the particular object further based on the determined angle and the reflectivity relationship between the first portion and the at least two additional portions.
4. The system of claim 1 , wherein the at least one processor is further configured to determine confidence scores for the reflectivity of the first portion and for the at least two additional portions and to classify the particular object further based on the determined confidence scores and the reflectivity relationship between the first portion and the at least two additional portions, wherein each of the confidence scores includes an indicator of a numerical value representing a level of confidence for a determined reflectivity.
5. The system of claim 1 , wherein the at least one processor is further configured to determine distances of the first portion and the at least two additional portions of the particular object, and to account for the determined distances when classifying the particular object.
6. The system of claim 1 , wherein the at least one processor is further configured to:
determine a reflectivity fingerprint of the particular object based on the reflectivity of the detected portions;
access memory that stores a plurality of indicators of fingerprints of various objects;
compare the reflectivity fingerprint of the particular object with the indicators of fingerprints of various objects stored in memory to identify a match; and
determine a type of the vehicle based on the identified match.
7. The system of claim 1 , wherein detecting the portions of the particular object includes analyzing time of flight information included in the point cloud information.
8. The system of claim 1 , wherein the point cloud information includes a point cloud map.
9. The system of claim 1 , wherein the system further comprises at least one memory configured to store classification information for classifying a plurality of objects, and wherein the at least one processor is further configured to:
receive a plurality of detection results associated with light detection and ranging (LIDAR) detection results, each detection result including location information, and further information indicative of at least two of the following detection characteristics:
object surface reflectivity;
temporal spreading of a signal reflected from the object;
object surface physical composition;
ambient illumination measured at a LIDAR dead time;
difference in detection information from a previous frame; and
confidence level associated with another detection characteristic, wherein the confidence level includes an indicator of a level of confidence for a detected characteristic;
access the classification information; and
based on the classification information and the detection results, classify an object in the vehicle's surroundings.
10. A non-transitory computer-readable storage medium storing program instructions executable by at least one processor to perform a method, the method comprising:
detecting, based on point cloud information, portions of a particular object;
determining, based on the detected portions, at least a first portion having a first reflectivity corresponding to a license plate, and at least two additional spaced-apart portions corresponding to locations on the particular object other than a location of the first portion, and wherein the at least two additional portions have reflectivity substantially lower than the first reflectivity; and
classifying the particular object as a vehicle, based on a spatial relationship and a reflectivity relationship between the first portion and the at least two additional portions.
11. The non-transitory computer-readable storage medium of claim 10 , the method further comprising:
determining a distance to the particular object; and
classifying the particular object based on the determined distance and the spatial relationship between the first portion and the at least two additional portions.
12. The non-transitory computer-readable storage medium of claim 10 , the method further comprising:
determining an angle of at least one surface associated with the particular object; and
classifying the particular object further based on the determined angle and the reflectivity relationship between the first portion and the at least two additional portions.
13. The non-transitory computer-readable storage medium of claim 10 , the method further comprising:
determining confidence scores for the reflectivity of the first portion and for the at least two additional portions; and
classifying the particular object further based on the determined confidence scores and the reflectivity relationship between the first portion and the at least two additional portions, wherein each of the confidence scores includes an indicator of a numerical value representing a level of confidence for a determined reflectivity.
14. The non-transitory computer-readable storage medium of claim 10 , the method further comprising:
determining distances of the first portion and the at least two additional portions of the particular object; and
accounting for the determined distances when classifying the particular object.
15. The non-transitory computer-readable storage medium of claim 10 , the method further comprising:
determining a reflectivity fingerprint of the particular object based on the reflectivity of the detected portions;
accessing memory that stores a plurality of indicators of fingerprints of various objects;
comparing the reflectivity fingerprint of the particular object with the indicators of fingerprints of various objects stored in memory to identify a match; and
determining a type of the vehicle based on the identified match.
16. The non-transitory computer-readable storage medium of claim 10 , wherein detecting the portions of the particular object includes analyzing time of flight information included in the point cloud information.
17. The non-transitory computer-readable storage medium of claim 10 , wherein the point cloud information includes a point cloud map.
18. The non-transitory computer-readable storage medium of claim 10 , the method further comprising:
receiving a plurality of detection results associated with light detection and ranging (LIDAR) detection results, each detection result including location information, and further information indicative of at least two of the following detection characteristics:
object surface reflectivity;
temporal spreading of a signal reflected from the object;
object surface physical composition;
ambient illumination measured at a LIDAR dead time;
difference in detection information from a previous frame; and
confidence level associated with another detection characteristic, wherein the confidence level includes an indicator of a level of confidence for a detected characteristic;
access classification information for classifying a plurality of objects, the classification information being stored in at least one memory; and
based on the classification information and the detection results, classify an object in the vehicle's surroundings.
19. A method, the method comprising:
detecting, based on point cloud information, portions of a particular object;
determining, based on the detected portions, at least a first portion having a first reflectivity corresponding to a license plate, and at least two additional spaced-apart portions corresponding to locations on the particular object other than a location of the first portion, and wherein the at least two additional portions have reflectivity substantially lower than the first reflectivity; and
classifying the particular object as a vehicle, based on a spatial relationship and a reflectivity relationship between the first portion and the at least two additional portions.
20. The system of claim 1 , wherein, in detecting the portions of the particular object, the at least one processor is configured to detect, based on the point cloud information, portions of the particular object that have a substantially same distance from a light source.