System and method for merging clusters
View Patent ↗A LiDAR point cloud that includes two candidate clusters for merging is received. At a first phase, a distance between the two clusters is determined. If the distance is greater than a threshold, the candidate clusters are not merged. Otherwise, an additional point cloud is received for each cluster at different times. A motion characteristic is determined for each cluster. If the motion characteristic for each cluster is close (indicating that the objects are moving at the same speed), then the clusters are merged. Otherwise the clusters are not merged. The motion characteristic for a cluster can be determined by performing an alignment operation using the point cloud received for the cluster, and using the error associated with the alignment operation as the motion characteristic for the cluster. The decision to merge clusters is based on raw point cloud data, which can take place early in the tracking cycle.
1. A system for merging clusters comprising:
one or more processors;
a memory communicably coupled to the one or more processors and storing:
a distance module including instructions that when executed by the one or more processors cause the one or more processors to:
receive, at a first time, a first point cloud for a first cluster and a second cluster, wherein the first point cloud includes a first plurality of points for the first cluster and a first plurality of points for the second cluster; and
determine a distance between the first cluster and the second cluster; and
a motion module including instructions that when executed by the one or more processors cause the one or more processors to:
in response to a determination that the distance satisfies a first threshold:
receive, at a second time, a second point cloud for the first cluster and the second cluster, wherein the second point cloud includes a second plurality of points for the first cluster and a second plurality of points for the second cluster;
receive, at a third time, a third point cloud for the first cluster and the second cluster, wherein the third point cloud includes a third plurality of points for the first cluster and a third plurality of points for the second cluster;
determine, based on the first plurality of points for the first cluster, the second plurality of points for the first cluster, and the third plurality of points for the first cluster, a first motion characteristic of the first cluster; and
determine, based on the first plurality of points for the second cluster, the second plurality of points for the second cluster, and the third plurality of points for the second cluster, a second motion characteristic of the second cluster; and
in response to a determination that the first motion characteristic and the second motion characteristic satisfy a second threshold:
determine to merge the first cluster and the second cluster.
2. The system of claim 1 , wherein the distance module further includes instructions that when executed by the one or more processors cause the one or more processors to:
in response to a determination that the distance does not satisfy the first threshold:
determine not to merge the first cluster and the second cluster.
3. The system of claim 1 , wherein the motion module further includes instructions that when executed by the one or more processors cause the one or more processors to:
in response to a determination that the first motion characteristic and the second motion characteristic do not satisfy the second threshold:
determine not to merge the first cluster and the second cluster.
4. The system of claim 1 , wherein the instructions to determine the first motion characteristic for the first cluster comprise:
instructions to perform a first alignment operation using the first plurality of points of the first cluster, the second plurality of points of the first cluster, and the third plurality of points of the first cluster;
instructions to determine, based on a result of the first alignment operation, one or more of a rotation value, a transformation value, and or an error value for the first cluster; and
instructions to determine the first motion characteristic using the one or more of the rotation value, the transformation value, or the error value determined for the first cluster.
5. The system of claim 4 , wherein the instructions to determine the second motion characteristic for the second cluster comprises:
instructions to perform a second alignment operation using the first plurality of points of the second cluster, the second plurality of points of the second cluster, and the third plurality of points of the second cluster;
instructions to determine, based on a result of the second alignment operation, one or more of a rotation value, a transformation value, or an error value for the second cluster; and
instructions to determine the second motion characteristic using the one or more of the rotation value, the transformation value, or the error value determined for the second cluster.
6. The system of claim 4 , wherein instructions to perform the first alignment operation use a trimmed iterative closest point algorithm.
7. The system of claim 1 , wherein the first point cloud is generated by a LiDAR sensor.
8. The system of claim 1 , wherein the distance module further includes instructions to select the first threshold based on a size of a vehicle.
9. The system of claim 1 wherein a boundary around the first plurality of points for the first cluster lacks a point of intersection with a boundary around the first plurality of points for the second cluster.
10. A method for merging clusters comprising:
receiving, at a first time, a first point cloud for a first cluster and a second cluster, wherein the first point cloud includes a first plurality of points for the first cluster and a first plurality of points for the second cluster;
determining a distance between the first cluster and the second cluster;
in response to a determination that the distance satisfies a first threshold:
receiving, at a second time, a second point cloud for the first cluster and the second cluster, wherein the second point cloud includes a second plurality of points for the first cluster and a second plurality of points for the second cluster;
receiving, at a third time, a third point cloud for the first cluster and the second cluster, wherein the third point cloud includes a third plurality of points for the first cluster and a third plurality of points for the second cluster;
determining, based on the first plurality of points for the first cluster, the second plurality of points for the first cluster, and the third plurality of points for the first cluster, a first motion characteristic for the first cluster; and
determining, based on the first plurality of points for the second cluster, the second plurality of points for the second cluster, and the third plurality of points for the cluster, a second motion characteristic for the second cluster; and
in response to a determination that the first motion characteristic and the second motion characteristic satisfy a second threshold:
determining to merge the first cluster and the second cluster.
11. The method of claim 10 , further comprising:
in response to a determination that the distance does not satisfy the first threshold:
determining not to merge the first cluster and the second cluster.
12. The method of claim 10 , further comprising:
in response to a determination that the first motion characteristic and the second motion characteristic do not satisfy the second threshold:
determining not to merge the first cluster and the second cluster.
13. The method of claim 10 , wherein the determining the first motion characteristic for the first cluster comprises:
performing a first alignment operation using the first plurality of points of the first cluster, the second plurality of points of the first cluster, and the third plurality of points of the first cluster;
determining, based on a result of the first alignment operation, one or more of a rotation value, a transformation value, or an error value for the first cluster; and
determining the first motion characteristic using the one or more of the rotation value, the transformation value, or the error value determined for the first cluster.
14. The method of claim 13 , wherein the determining the second motion characteristic for the second cluster comprises:
performing a second alignment operation using the first plurality of points of the second cluster, the second plurality of points of the second cluster, and the third plurality of points of the second cluster;
determining, based on a result of the second alignment operation, one or more of a rotation value, a transformation value, or an error value for the second cluster; and
determining the second motion characteristic using the one or more of the rotation value, the transformation value, or the error value determined for the second cluster.
15. The method of claim 13 , wherein the performing the first alignment operation uses a trimmed iterative closest point algorithm.
16. The method of claim 10 , wherein the first point cloud is generated by a LiDAR sensor.
17. The method of claim 10 , further comprising selecting the first threshold based on a size of a vehicle.
18. A method for merging clusters comprising:
receiving, at a first time, a first point cloud for a first cluster and a second cluster, wherein the first point cloud includes a first plurality of points for the first cluster and a first plurality of points for the second cluster;
determining a distance between the first cluster and the second cluster;
in response to a determination that the distance satisfies a first threshold:
receiving, at a second time, a second point cloud for the first cluster and the second cluster, wherein the second point cloud includes a second plurality of points for the first cluster and a second plurality of points for the second cluster;
determining, based on the first plurality of points for the first cluster and the second plurality of points for the first cluster, a first motion characteristic for the first cluster; and
determining, based on the first plurality of points for the second cluster and the second plurality of points for the second cluster, a second motion characteristic for the second cluster; and
in response to a determination that the first motion characteristic and the second motion characteristic satisfy a second threshold:
determining to merge the first cluster and the second cluster.
19. The method of claim 18 , further comprising:
in response to a determination that the distance does not satisfy the first threshold:
determining not to merge the first cluster and the second cluster.
20. The method of claim 18 , further comprising:
in response to a determination that the first motion characteristic and the second motion characteristic do not satisfy the second threshold:
determining not to merge the first cluster and the second cluster.
21. The method of claim 18 , wherein the first point cloud is generated by a LiDAR sensor.