Using image and laser constraints to obtain consistent and improved pose estimates in vehicle pose databases

• US 8,437,501 B1
• Filed: 08/03/2012
• Issued: 05/07/2013
• Est. Priority Date: 09/14/2010
• Status: Active Grant

• Current Assignee: Google LLC (Alphabet Inc.)
• Original Assignee: Google Inc. (Alphabet Inc.)
• Inventors: Anguelov, Dragomir, Agarwal, Sameer, Frueh, Christian
• Primary Examiner(s): Seth, Manav
• Application Number:
  US13/566,199
• Time in Patent Office: 277 Days
• Field of Search: 382100-322
• US Class Current: 382/100
• CPC Class Codes: G06T 2207/30241 Trajectory
• CPC Class Codes: G06T 7/73 using feature-based methods
• CPC Class Codes: G06V 20/10 Terrestrial scenes scenes under surveillance with static cameras G06V20/52; scenes perceived from the exterior of a vehicle G06V20/56; scenes perceived from the interior of a vehicle G06V20/59

Abstract

A system for pose generation consisting of a trajectory system, a pose generation system, an intersection extractor, an object identifier, a constraint generator, and a posegraph solver. The trajectory system identifies a number of trajectories based on input positional data of a bounded area. The pose generation system generates one or more poses based on the trajectories. The intersection extractor identifies one or more possible intersections in the one or more poses. The object identifier identifies an object pair for each possible intersection that represents two positional points at each possible intersection. The constraint generator computes and applies one or more intersection constraints to generate an energy value for each object pair based on their geometric relationship. The posegraph solver then minimizes a total energy value by modifying one or more poses and then generates an improved set of pose trajectories based on the modified one or more poses.

Claims

1. A computer-implemented method for pose generation comprising:

identifying a first bounded area to be analyzed;

obtaining positional input data of the first bounded area;

identifying a single trajectory, comprising a first portion and a second portion, based on the positional input data;

generating one or more poses based on the first portion and the second portion of the single trajectory;

defining one or more possible intersections for each pose;

identifying a first object pair at a first possible intersection representing two positional points at the first possible intersection;

computing one or more intersection constraints;

applying the one or more intersection constraints to generate an energy value based on a geometric relationship of the first object pair; and

modifying the one or more poses to minimize a total energy value.

2. The method of claim 1, further comprising generating an improved set of pose trajectories based on the modified one or more poses.

3. The method of claim 1, wherein the first portion is separated from the second portion by a difference in time.

4. The method of claim 1, wherein one of the one or more intersection constraints is based on vision or laser sensor data.

5. The method of claim 1, wherein one of the one or more intersection constraints is based on global positioning system data.

6. The method of claim 1, wherein one of the one or more intersection constraints is based on odometry data.

7. The method of claim 1, further comprising:

identifying a second object pair at a second possible intersection representing two positional points at the second possible intersection; and

applying the one or more intersection constraints to generate an energy value based on a geometric relationship of the second object pair.

8. The method of claim 1, further comprising:

identifying a second bounded area to be analyzed; and

obtaining positional input data of the second bounded area,

wherein identified trajectories in the first and second bounded areas are non-overlapping.

9. The method of claim 8, further comprising:

identifying a third bounded area to be analyzed; and

obtaining positional input data of the second bounded area,

wherein identified trajectories in the first, second, and third bounded areas are non-overlapping, and

wherein the third bounded area is not adjacent to the first bounded area.

10. The method of claim 1, wherein the obtaining positional input data is based on the use of a vehicle.

11. A system for pose generation comprising:

a trajectory system configured to identify a single trajectory, comprising a first portion and a second portion, based on the positional input data of a first bounded area;

a pose generation system configured to generate one or more poses based on the first portion and the second portion of the single trajectory;

an intersection extractor configured to identify one or more possible intersections in the one or more poses;

an object identifier configured to identify a first object pair at a first possible intersection representing two positional points at the first possible intersection;

a constraint generator configured to compute one or more intersection constraints and apply the one or more intersection constraints to generate an energy value based on a geometric relationship of the first object pair; and

a posegraph solver configured to minimize a total energy value by modifying the one or more poses.

12. The system of claim 11, wherein the posegraph solver generates an improved set of pose trajectories based on the modified one or more poses.

13. The system of claim 11, the first portion of the single trajectory occurred at a different time than the second portion of the single trajectory.

14. The system of claim 11, wherein one of the one or more intersection constraints is based on vision or laser sensor data.

15. The system of claim 11, wherein one of the one or more intersection constraints is based on global positioning system data.

16. The system of claim 11, wherein one of the one or more intersection constraints is based on odometry data.

17. The system of claim 11, further comprising the object identifier configured to identify a second object pair at a second possible intersection representing two positional points at the second possible intersection, and the constraint generator configured to generate an energy value based on a geometric relationship of the second object pair.

18. The system of claim 11, wherein a storage system is configured to contain positional input data for a second bounded area to be analyzed, and wherein the trajectory system identifies trajectories in the first and second bounded areas that are non-overlapping.

19. The system of claim 18, wherein the storage system is configured to contain positional input data for a third bounded area to be analyzed, wherein the trajectory system identifies trajectories in the first, second, and third bounded areas are non-overlapping, and wherein the third bounded area is not adjacent to the first bounded area.

20. The system of claim 11, further comprising a vehicle configured to collect positional input data.