Systems and methods for scenario dependent trajectory scoring

Patent metadata

Field	Value
Publication number	US-12337868-B2
Application number	US-202117152894-A
Country	US
Kind code	B2
Filing date	Jan 20, 2021
Priority date	Jan 20, 2021
Publication date	Jun 24, 2025
Grant date	Jun 24, 2025

Abstract

Systems and methods for operating an autonomous vehicle. The methods comprising: obtaining one or more candidate vehicle trajectories for the autonomous vehicle and context information defining a state of an environment surrounding the autonomous vehicle; assigning class(es) to a scenario specified by the context information and a first candidate vehicle trajectory; generating a first quality score for the first candidate vehicle trajectory using scoring function(s) selected based on the assigned class(es); select a candidate vehicle trajectory based on the first quality score associated with the first candidate vehicle trajectory and second quality score(s) associated with at least one second candidate vehicle trajectory; and causing the autonomous vehicle to perform autonomous driving operations using the selected candidate vehicle trajectory.

First claim

What is claimed is:

1. A method for operating an autonomous vehicle, comprising:
obtaining, by a computing device, one or more candidate vehicle trajectories for the autonomous vehicle and context information defining a state of an environment surrounding the autonomous vehicle;
classifying, by the computing device, a context into at least one scenario class of a plurality of different scenario classes based on the context information and a first candidate vehicle trajectory of the candidate vehicle trajectories, the context being classified into a first scenario class and a second scenario class;
selecting, by the computing device for a scenario class among the at least one scenario class into which the context was classified, a set of scoring functions from a plurality of different sets of scoring functions for the first scenario class and one scoring function for the second scenario class, the plurality of different sets of scoring functions being respectively associated with the plurality of different scenario classes, and the set of scoring functions including a plurality of scoring functions for the scenario class;
generating, by the computing device, a plurality of scores using the set of scoring functions for the first scenario class;
combining, by the computing device, the plurality of scores by performing a summation of the plurality of scores to obtain a first scenario score for the first scenario class;
generating, by the computing device, a second scenario score for the second scenario class using the one scoring function;
generating, by the computing device, a first quality score for the first candidate vehicle trajectory by combining the first scenario score and the second scenario score;
selecting, by the computing device, a candidate vehicle trajectory from the candidate vehicle trajectories based on the first quality score associated with the first candidate vehicle trajectory and a second quality score associated with a second candidate vehicle trajectory; and
causing, by the computing device, the autonomous vehicle to perform autonomous driving operations using the selected candidate vehicle trajectory.

2. The method according to claim 1 , wherein the context information comprises at least one of intersection information, object related information, a road map, traffic information, and environmental information.

3. The method according to claim 1 , further comprising performing operations, by the computing device, to (i) generate a feature vector using the context information and the first candidate vehicle trajectory of the candidate vehicle trajectories and (ii) use the feature vector to generate the quality score for the first candidate vehicle trajectory.

4. The method according to claim 1 , wherein
the context is classified into a first scenario class and a second scenario class,
the set of scoring functions is selected for the first scenario class,
the method further comprising selecting one scoring function from among of a plurality of scoring functions based on the second scenario class, wherein:
a first scenario score is generated using the set of scoring functions for the first scenario class and a second scenario score is generated using the one scoring function for the second scenario class, and
the first scenario score and the second scenario score are combined together to produce the first quality score for the first candidate vehicle trajectory.

5. The method according to claim 1 , wherein the plurality of different scenarios classes further comprise at least one of a passing scenario class, an acceleration scenario class, a deceleration scenario class, a stationary scenario class, a forward driving scenario class, a reverse driving scenario class, and/or a passenger pick-up scenario class.

6. The method according to claim 1 , wherein the candidate vehicle trajectory is selected from the candidate vehicle trajectories by comparing the first and second quality scores to each other or to a threshold value.

7. The method according to claim 1 , wherein the autonomous driving operations comprise causing the autonomous vehicle to follow the selected candidate vehicle trajectory.

8. A system, comprising:
a processor;
a non-transitory computer-readable storage medium comprising programming instructions that are configured to cause the processor to implement a method for operating an autonomous vehicle, wherein the programming instructions comprise instructions to:
obtain one or more candidate vehicle trajectories for the autonomous vehicle and context information defining a state of an environment surrounding the autonomous vehicle;
classify a context into at least one scenario class of a plurality of different scenario classes based on the context information and a first candidate vehicle trajectory of the candidate vehicle trajectories, the context being classified into a first scenario class and a second scenario class;
select a set of scoring functions from a plurality of different sets of scoring functions for the first scenario class and one scoring function for the second scenario class, the plurality of different sets of scoring functions being respectively associated with the plurality of different scenario classes, and the set of scoring functions including a plurality of scoring functions;
generate a plurality of scores using the set of scoring functions for the first scenario class;
combine the plurality of scores for the scenario class to obtain a first scenario score for the first scenario class by performing a summation of the plurality of scores;
generate a second scenario score for the second scenario class using the one scoring function;
generate at least one first quality score for the first candidate vehicle trajectory by combining the first scenario score and the second scenario score;
select candidate vehicle trajectory from the candidate vehicle trajectories based on the first quality score associated with the first candidate vehicle trajectory and a second quality score associated with at least one second candidate vehicle trajectory; and
cause the autonomous vehicle to perform autonomous driving operations using the selected candidate vehicle trajectory.

9. The system according to claim 8 , wherein the context information comprises at least one of intersection information, object related information, a road map, traffic information, and environmental information.

10. The system according to claim 8 , wherein the programming instructions further comprise instructions to (i) generate a feature vector using the context information and the first candidate vehicle trajectory of the candidate vehicle trajectories and (ii) use the feature vector to generate the quality score for the first candidate vehicle trajectory.

11. The system according to claim 8 , wherein the plurality of different scenario classes comprise at least one of a left turn scenario class, a right turn scenario, a passing scenario class, a driving scenario class, an acceleration scenario class, a deceleration scenario class, a stationary scenario class, a forward driving scenario class, a reverse driving scenario class, and/or a passenger pick-up scenario class.

12. The system according to claim 8 , wherein the candidate vehicle trajectory is selected from the candidate vehicle trajectories by comparing the first and second quality scores to each other or to a threshold value.

13. The system according to claim 8 , wherein the autonomous driving operations comprise causing the autonomous vehicle to follow the selected candidate vehicle trajectory.

14. A non-transitory computer-readable medium that stores

Assignees

• Ford Global Tech Llc

Inventors

• Wang Yuh-Shyang
• Cunningham Christopher
• Honka Nathaniel
• Mirkovic Damir

Classifications

• B60W40/06

  Road conditions · CPC title

• B60W40/04

  Traffic conditions · CPC title

• B60W2554/404

  Characteristics · CPC title

• B60W30/09

  Taking automatic action to avoid collision, e.g. braking and steering · CPC title

• B60W60/0015

  specially adapted for safety · CPC title