Verifying predicted trajectories using a grid-based approach

Patent metadata

Field	Value
Publication number	US-12204333-B2
Application number	US-202117191491-A
Country	US
Kind code	B2
Filing date	Mar 3, 2021
Priority date	Nov 9, 2018
Publication date	Jan 21, 2025
Grant date	Jan 21, 2025

Abstract

Aspects of the disclosure provide for controlling a vehicle in an autonomous driving mode. For instance, sensor data for an object as well as a plurality of predicted trajectories may be received. Each predicted trajectory may represent a plurality of possible future locations for the object. A grid including a plurality of cells, each being associated with a geographic area, may be generated. Probabilities that the object will enter the geographic area associated with each of the plurality of cells over a period of time into the future may be determined based on the sensor data in order to generate a heat map. One or more of the plurality of predicted trajectories may be compared to the heat map. The vehicle may be controlled in the autonomous driving mode based on the comparison.

First claim

The invention claimed is:

1. A method, comprising:
receiving, by one or more processors, sensor data for an object detected in an environment of a vehicle;
receiving, by the one or more processors, a plurality of predicted trajectories for the object, each predicted trajectory representing a plurality of possible future locations for the object;
generating, by the one or more processors, a grid of cells around the object, the grid of cells having a boundary associated with motion of the object, each cell of the grid of cells representing a respective portion of a geographic area around the object;
determining, by the one or more processors based on the sensor data, for each cell of the grid of cells, a respective probability that the object, over a period of time into the future, will enter the respective portion of the geographic area;
identifying, by the one or more processors, one or more cells of the grid of cells that overlap with one or more of the plurality of predicted trajectories; and
controlling, by the one or more processors, the vehicle in an autonomous driving mode based on the respective probabilities for the one or more identified cells.

2. The method of claim 1 , further comprising, prior to identifying the one or more cells, identifying the one or more of the plurality of predicted trajectories based on respective trajectory probability values associated with the plurality of predicted trajectories and a threshold probability value.

3. The method of claim 2 , wherein:
the threshold probability value is a threshold minimum probability value,
identifying the one or more of the plurality of predicted trajectories includes determining, by the one or more processors, whether each of the respective trajectory probability values is at least the threshold minimum probability value, and
the respective trajectory probability values of the one or more of the plurality of predicted are determined to be at least the threshold minimum probability value.

4. The method of claim 1 , further comprising, generating, by the one or more processors, the plurality of predicted trajectories based on the sensor data.

5. The method of claim 1 , wherein the object is another vehicle different from the vehicle.

6. The method of claim 1 , further comprising:
determining, by the one or more processors based on the sensor data, whether the object is of a particular type of object; and
generating, by the one or more processors, the grid in response to determining that the object is of the particular type of object.

7. The method of claim 1 , further comprising:
determining, by the one or more processors based on the sensor data, whether a speed of the object is at least a threshold speed; and
generating, by the one or more processors, the grid in response to determining that the speed of the object is at least the threshold speed.

8. The method of claim 1 , further comprising
receiving, by the one or more processors, sensor data for another object detected in the environment of the vehicle;
receiving, by the one or more processors, a plurality of predicted trajectories for the other object, each predicted trajectory representing a plurality of possible future locations for the other object; and
generating, by the one or more processors, another grid of cells around the other object, a dimension of the other grid of cells being based on a speed of the other object, each cell of the other grid of cells representing a respective portion of a geographic area around the other object.

9. The method of claim 8 , further comprising:
determining, by the one or more processors based on the sensor data for the other object, for each cell of the other grid of cells, a respective probability that the other object, over another period of time into the future, will enter the respective portion of the geographic area;
identifying, by the one or more processors, one or more cells of the other grid of cells that overlap with one or more of the plurality of predicted trajectories for the other object; and
controlling, by the one or more processors, the vehicle in the autonomous driving mode based on the respective probabilities for the one or more identified cells of the other grid of cells.

10. The method of claim 1 , further comprising, assessing, by the one or more processors, validity of the one or more of the plurality of predicted trajectories based on the respective probabilities for the one or more identified cells.

11. The method of claim 1 , further comprising, based on the respective probabilities for the one or more identified cells, generating, by the one or more processors, a new predicted trajectory based on one or more of the respective probabilities for the grid of cells, and wherein the controlling is further based on the new predicted trajectory.

12. The method of claim 11 , wherein generating the new predicted trajectory includes examining, by the one or more processors, border cells of the grid of cells and taking an average of border cell locations weighted by the respective probabilities for the border cells.

13. The method of claim 12 , wherein generating the new predicted trajectory further includes generating, by the one or more processors, a constant curvature trajectory using the average and a location of the object.

14. The method of claim 11 , wherein generating the new predicted trajectory includes taking, by the one or more processors, an average of cell locations for two or more cells of the grid of cells weighted by the respective probabilities for the two or more cells.

15. The method of claim 14 , wherein generating the new predicted trajectory further includes generating, by the one or more processors, a constant curvature trajectory using the average and a location of the object.

16. The method of claim 1 , further comprising:
identifying, by the one or more processors, a first set of cells of the grid of cells based on the respective probabilities for the first set of cells being at least a threshold minimum probability value; and
determining, by the one or more processors, whether the one or more of the plurality of predicted trajectories overlap the first set of cells.

17. The method of claim 16 , further comprising, for a given predicted trajectory of the one or more of the plurality of predicted trajectories:
identifying, by the one or more processors, a second set of cells of the grid of cells closest to points of the given predicted trajectory; and
determining, by the one or more processors, whether the first set of cells includes one or more of the second set of cells.

18. The method of claim 16 , further comprising, for a given predicted trajectory of the one or more of the plurality of predicted trajectories:
identifying, by the one or more processors, a border cell of the grid of cells that overlaps the given predicted trajectory; and
determining, by the one or more processors, whether the first set of cells includes the border cell.

19. The method of claim 1 , further comprising, determining, by the one or more processors, whether the respective probabilities for the one or more identified cells are at least a threshold minimum probability value.

20. The method of claim 19 , further comprising, responsive to determining that none of the respective probabilities for the one or more identified cells is at least the threshold minimum probability value, flagging, by the one or more processors, the one or more of the plurality of predicted trajectories as anomalous.

Assignees

• Waymo Llc

Inventors

• Refaat Khaled
• Ding Kai

Classifications

• G06N3/08

  Learning methods · CPC title

• G05D1/617

  Safety or protection, e.g. defining protection zones around obstacles or avoiding hazards (arrangements for controlling the position or course of two or more vehicles for avoiding collisions therebetween G05D1/693; arrangements for reacting to or preventing system or operator failure G05D1/80) · CPC title

• G05D1/81

  Handing over between on-board automatic and on-board manual control · CPC title

• G05D1/65

  Following a desired speed profile · CPC title

• G05D1/0221

  involving a learning process · CPC title