| Sonstiges: |
- Nachgewiesen in: USPTO Patent Applications
- Sprachen: English
- Document Number: 20230186494
- Publication Date: June 15, 2023
- Appl. No: 17/549516
- Application Filed: December 13, 2021
- Claim: 1. A system comprising: a LIDAR sensor; and one or more processors configured to: receive a first set of LIDAR data points associated with a first position and orientation of a first vehicle in an environment; receive a second set of LIDAR data points associated with a second position and orientation of a second vehicle in the environment; associate the first set of LIDAR data points with a first voxel grid comprising a plurality of first voxels, a first voxel associated with a first sub-set of the first set of LIDAR data points; determine, based at least in part on the associated first sub-set of the first set of LIDAR data points, respective first characteristics of the first voxels; associate the second set of LIDAR data points with a second voxel grid comprising a plurality of second voxels, a second voxel associated with a corresponding second sub-set of the second set of LIDAR data points; determine, based at least in part on the associated second sub-set of the second set of LIDAR data points, respective second characteristics of the second voxels; determine a pair of first and second voxels that are at least partially aligned; determine, based at least in part on a comparison between the first characteristic of the first voxel of the pair and the second characteristic of the second voxel of the pair, a similarity metric; determine, based at least in part on the similarity metric, whether a point cloud registration between the first and second sets of LIDAR data points is invalid; assign, based at least in part on the determination as to whether the point cloud registration between the first and second sets of LIDAR data points is invalid, a weight to the point cloud registration; determine, based at least in part on the weight, the first set of LIDAR data points, and the second set of LIDAR data points, a map; and transmit the map to an additional vehicle configured to be controlled based at least in part on the map.
- Claim: 2. A system as claimed in claim 1, wherein: the first characteristics comprise first classification labels of the respective first sub-set of LIDAR data points associated with the first voxel; the second characteristics comprise second classification labels of the respective second sub-set of LIDAR data points associated with the second voxel; and the one or more processors are configured to determine, based at least in part on dissimilarities between the first and second classification labels of the first and second voxels of the pair, the similarity metric.
- Claim: 3. A system as claimed in claim 2, wherein the one or more processors are configured to: for the pair of first and second voxels, determine, based at least in part on the first classification labels, a first distribution of classification labels; determine, based at least in part on the second classification labels, a second distribution of classification labels; and determine, based at least in part on a comparison between the first and second distributions of classification labels, the similarity metric.
- Claim: 4. A system as claimed in claim 1, wherein: the first and second characteristics comprise classifications as to whether the respective first and second voxels are free space or occupied; and the one or more processors are further configured to determine whether the respective first and second voxels are free space or occupied based on at least one of a ray casting and a comparison of a number of LIDAR data points in the first and second sub-sets of data points to a threshold value.
- Claim: 5. A method comprising: obtaining first and second sets of data points corresponding to respective first and second poses of a vehicle in an environment; determining first and second voxel grids corresponding to the first and second sets of data points; determining a pair of voxels between the first and second voxel grids, the pair of voxels at least partially aligned with one another; determining, based at least in part on a comparison between a first characteristic of a first voxel of the pair and a second characteristic of a second voxel of the pair, a similarity metric; and determining, based at least in part on the similarity metrics, whether a point cloud registration between the first and second sets of data points is invalid.
- Claim: 6. A method as claimed in claim 5, wherein: the first characteristic comprises first classification labels of a respective first sub-set of data points of the first set of data points associated with the first voxel; the second characteristic comprises second classification labels of a respective second sub-set of data points of the second set of data points associated with the second voxel; and the method comprises determining, based at least in part on dissimilarities between the first and second classification labels of the first and second voxels of the pair, the similarity metric.
- Claim: 7. A method as claimed in claim 6, wherein the method comprises; for the pair of voxels, determining, based at least in part on the first classification labels, a first distribution of classification labels; determining, based at least in part on the second classification labels, a second distribution of classification labels; and determining, based at least in part on a comparison between the first and second distributions, the similarity metric.
- Claim: 8. A method as claimed in claim 5, wherein the method comprises: determining a plurality of pairs of first and second voxels between the first and second voxel grids; determining, for the plurality of pairs and based at least in part on an associated plurality of characteristics, a plurality of similarity metrics; and determining, based at least in part on the plurality of similarity metrics, a distribution of similarity metrics, wherein determining whether the point cloud registration is invalid is based at least in part on the distribution of similarity metrics.
- Claim: 9. A method as claimed in claim 8, wherein determining whether the point cloud registration is invalid comprises determining an expected value of the distribution of similarity metrics, and comparing the expected value to a threshold.
- Claim: 10. A method as claimed in claim 5, wherein: the first and second characteristics comprise classifications as to whether the respective first and second voxels are free space or occupied; and the method comprises determining whether the respective first and second voxels are free space or occupied based on a comparison of a number of LIDAR data points in the first and second sub-sets of data points to a threshold value.
- Claim: 11. A method as claimed in claim 5, wherein: the first and second characteristics comprise classifications as to whether the respective first and second voxels are free space or occupied; and the method comprises determining whether the respective first and second voxels are free space or occupied based on a ray casting.
- Claim: 12. A method as claimed in claim 5, wherein: one or more of the first or second characteristic comprises a classification of: free space, occupied space, building, drivable surface, road, vegetation, static object or dynamic object.
- Claim: 13. A method as claimed in claim 5, wherein the method comprises: determining, based at least in part on one or more of a position, orientation or environmental characteristic, whether the point cloud registration between the first and second sets of data points is invalid.
- Claim: 14. A method as claimed in claim 5, wherein the method comprises disregarding, from the first and second sets of data points, data points indicative of dynamic objects in an environment of the vehicle at the respective first and second poses of the vehicle.
- Claim: 15. A method as claimed in claim 5, wherein the first and second sets of data points comprise data from one or more of LIDARs, cameras, or radars.
- Claim: 16. A method as claimed in claim 5, wherein the method comprises: determining, based at least in part on whether the point cloud registration between the first and second sets of data points is invalid, a weight assigned to the point cloud registration; and determining, based at least in part on the first and second sets of data points and the weight assigned to the point cloud registration between the first and second sets of data points, a map of an environment of the vehicle.
- Claim: 17. A method as claimed in claim 16, wherein the method comprises transmitting the map to an additional vehicle configured to be controlled based at least in part on the map.
- Claim: 18. One or more non-transitory computer-readable media storing instructions executable by one or more processors of a system, wherein the instructions, when executed, cause the one or more processors to perform operations comprising: obtaining first and second sets of data points corresponding to respective first and second poses of a vehicle in an environment; determining first and second voxel grids corresponding to the first and second sets of data points; determining a pair of voxels between the first and second voxel grids, the pair of voxels at least partially aligned with one another; determining, based at least in part on a comparison between a first characteristic of a first voxel of the pair and a second characteristic of a second voxel of the pair, a similarity metric; and determining, based at least in part on the similarity metrics, whether a point cloud registration between the first and second sets of data points is invalid.
- Claim: 19. One or more non-transitory computer-readable media as claimed in claim 18, wherein: the first characteristic comprises first classification labels of a respective first sub-set of data points of the first set of data points associated with the first voxel; the second characteristic comprises second classification labels of a respective second sub-set of data points of the second set of data points associated with the second voxel; and the instructions, when executed, cause the one or more processors to perform operations comprising determining, based at least in part on dissimilarities between the first and second classification labels of the first and second voxels of the pair, the similarity metric.
- Claim: 20. One or more non-transitory computer-readable media as claimed in claim 18, wherein: the first and second characteristics comprise classifications as to whether the respective first and second voxels are free space or occupied; and the one or more processors are further configured to determine whether the respective first and second voxels are free space or occupied based on at least one of a ray casting and a comparison of a number of LIDAR data points in the first and second sub-sets of data points to a threshold value.
- Current International Class: 06; 06; 06; 06
|
