Granted Patent B2

US 12,371,053 · App. 18/064,083 · Granted Jul 29, 2025

Inventors: Oliver F. Schwindt (Sunnyvale, CA); Stephan Reuter (Bavaria, DE); Christos Zalidis (Mountain View, CA); Tobias Berling (Sunnyvale, CA)

Assignee: Robert Bosch GmbH

B60W60/001B60W2554/40

Abstract

Claims (70)

1. A system for operating an autonomous vehicle, the system comprising:

a sensor configured to output sensor data representative of a detected dynamic object,

an electronic processor configured to:

receive a static occupancy grid, the static occupancy grid including a plurality of cells, a subset of the cells storing occupancy information and occupancy probabilities associated with a static object in the environment,

generate a dynamic object list, the dynamic object list including dynamic object information associated with the detected dynamic object and a respective position of the detected dynamic object in the environment,

superimpose a representation of the detected dynamic object on the static occupancy grid based on the dynamic object information,

generate a dynamic object perimeter based on a shape of the detected dynamic object as superimposed on the static occupancy grid,

determine, based on the dynamic object perimeter, an overlap of the detected dynamic object and the static object represented in the static occupancy grid,

determine a dynamic object existence probability associated with the detected dynamic object based at least on the overlap,

classify the detected dynamic object as a ghost dynamic object in response to determining the dynamic object existence probability is less than a threshold, and

control vehicle motion based on the classification of the detected dynamic object as a ghost dynamic object.

2. The system of claim 1 , wherein the electronic processor is further configured to, in response to the dynamic object existence probability being greater than the threshold, classify the static object as a ghost static object.

3. The system of claim 1 , wherein the electronic processor is configured to determine the dynamic object existence probability by

determining a modality of the sensor, and

determining the dynamic object existence probability based on the modality of the sensor.

4. The system of claim 1 , wherein the electronic processor is configured to determine the dynamic object existence probability by

determining a size of the detected dynamic object, and

determining the dynamic object existence probability based on the size of the detected dynamic object.

5. The system of claim 1 , wherein

the system includes a plurality of sensors configured to detect the dynamic object, and

the electronic processor is configured to determine the dynamic object existence probability by

determining a number of sensors that detected the dynamic object, and

determining the dynamic object existence probability based on the number of sensors that detected the dynamic object.

6. The system of claim 1 , wherein the electronic processor is configured to determine the dynamic object existence probability by

tracking the detected dynamic object over a period of time,

determining a plurality of dynamic object existence probabilities over the period of time, and

determining a historical dynamic object existence probability associated with the detected dynamic object based on the plurality of dynamic object existence probabilities.

7. The system for an autonomous vehicle of claim 1 , wherein the electronic processor is configured to determine the dynamic object existence probability by

determining a direction of motion of the detected dynamic object,

determining a plurality of dynamic object existence probabilities, each associated with a cell of the static occupancy grid included in the overlap, and

applying a weight to a subset of the plurality of dynamic object existence probabilities based on the direction of motion.

8. The system for an autonomous vehicle of claim 1 , wherein the electronic processor is configured to superimpose the dynamic object information on the static occupancy grid by

predicting or retrodicting the dynamic object information such that a time stamp associated with the dynamic object information matches a time stamp associated with the static occupancy grid, and

spatially transforming the dynamic object information.

9. The system for an autonomous vehicle of claim 1 , wherein the plurality of cells of the static occupancy grid are voxels, and the dynamic object perimeter is a 3D bounding box.

10. A method for operating an autonomous vehicle, the method comprising:

outputting, with a sensor, sensor data representative of a detected dynamic object;

receiving, with an electronic processor, a static occupancy grid, the static occupancy grid including a plurality of cells, a subset of the cells storing occupancy information and occupancy probabilities associated with a static object in the environment;

generating a dynamic object list, the dynamic object list including dynamic object information associated with the detected dynamic object and a respective position of the detected dynamic object in the environment;

superimposing a representation of the detected dynamic object on the static occupancy grid based on the dynamic object information;

generating a dynamic object perimeter based on a shape of the detected dynamic object as superimposed on the static occupancy grid;

determining, based on the dynamic object perimeter, an overlap of the detected dynamic object and the static object represented in the static occupancy grid;

determining a dynamic object existence probability associated with the detected dynamic object based at least on the overlap;

classifying the detected dynamic object as a ghost dynamic object in response to determining the dynamic object existence probability is less than a threshold; and

controlling vehicle motion based on the classification of the detected dynamic object as a ghost dynamic object.

11. The method of claim 10 , further comprising

in response to the dynamic object existence probability being greater than the threshold, classifying the static object as a ghost static object.

12. The method of claim 10 , wherein determining the dynamic object existence probability includes

determining a modality of the sensor, and

determining the dynamic object existence probability based on the modality of the sensor.

13. The method of claim 10 , wherein determining the dynamic object existence probability includes

determining a size of the detected dynamic object, and

determining the dynamic object existence probability based on the size of the detected dynamic object.

14. The method of claim 10 , further comprising

detecting the dynamic object with a plurality of sensors, and

wherein determining the dynamic object existence probability includes

determining a number of sensors that detected the dynamic object, and

determining the dynamic object existence probability based on the number of sensors that detected the dynamic object.

15. The method of claim 10 , wherein determining the dynamic object existence probability includes

tracking the detected dynamic object over a period of time,

determining a plurality of dynamic object existence probabilities over the period of time, and

determining a historical dynamic object existence probability associated with the detected dynamic object based on the plurality of dynamic object existence probabilities.

16. The method of claim 10 , wherein determining the dynamic object existence probability includes

determining a direction of motion of the detected dynamic object,

determining a plurality of dynamic object existence probabilities, each associated with a cell of the static occupancy grid included in the overlap, and

applying a weight to a subset of the plurality of dynamic object existence probabilities based on the direction of motion.

17. The method of claim 10 , wherein superimposing the dynamic object information on the static occupancy grid includes

predicting or retrodicting the dynamic object information such that a time stamp associated with the dynamic object information matches a time stamp associated with the static occupancy grid, and

spatially transforming the dynamic object information.

18. The method of claim 10 , wherein the plurality of cells of the static occupancy grid are voxels, and the dynamic object perimeter is a 3D bounding box.

Assignments (1)

Continuity (1)

References Cited (27)