Traffic boundary mapping

Patent metadata

Field	Value
Publication number	US-11874119-B2
Application number	US-202117145230-A
Country	US
Kind code	B2
Filing date	Jan 8, 2021
Priority date	Mar 23, 2018
Publication date	Jan 16, 2024
Grant date	Jan 16, 2024

Abstract

The present disclosure provides devices, systems and methods for mapping of traffic boundaries.

First claim

What is claimed is:

1. A method, comprising:
receiving visual data from a camera on a vehicle;
detecting a traffic boundary within the visual data;
selecting a first one or more cells of an occupancy grid based on a bird's eye view projection of the detected traffic boundary, wherein each cell of the occupancy grid corresponds to a position on a road relative to the vehicle;
incrementing a value of the first one or more cells;
receiving a portion of a traffic boundary map;
computing a cross-correlation between the occupancy grid and the portion of the traffic boundary map to produce at least one consistency value; and
localizing the vehicle within the traffic boundary map based on the at least one consistency value.

2. The method of claim 1 , further comprising:
selecting a second one or more cells of the occupancy grid, wherein the second one or more cells substantially surround the first one or more cells; and
decrementing a value of the second one or more cells.

3. The method of claim 1 , further comprising:
determining a position, with respect to the portion of the traffic boundary map, of a particle in a particle filter; wherein a consistency value of the at least one consistency value corresponds to the determined position of the particle.

4. The method of claim 3 , further comprising:
receiving inertial measurement data from a sensor on the vehicle; wherein the position of the particle is based at least in part on the inertial measurement data.

5. The method of claim 3 , further comprising:
receiving global positioning data from a sensor on the vehicle; wherein the position of the particle is based at least in part on the global positioning data.

6. The method of claim 3 , wherein a resampling weight for the particle is based on the determined consistency value corresponding to the particle.

7. The method of claim 1 , wherein the portion of the traffic boundary map comprises a spline, wherein the spline represents a previously mapped traffic boundary.

8. The method of claim 1 , wherein the portion of the traffic boundary map comprises a second occupancy grid, wherein values of cells of the second occupancy grid indicate locations of previously mapped traffic boundaries.

9. The method of claim 1 , wherein more than one consistency values are computed, each consistency value of the more than one consistency values corresponding to a spatial offset between the occupancy grid and the traffic boundary map, and each spatial offset corresponding to a candidate location of the vehicle within the traffic boundary map.

10. The method of claim 9 , further comprising:
normalizing the more than one consistency values to produce a spatial probability distribution, wherein the spatial probability distribution quantifies a probability of the vehicle being located at each of a plurality of candidate locations within the traffic boundary map.

11. The method of claim 9 , wherein localizing the vehicle comprises:
selecting a candidate location of the plurality of candidate locations having a highest consistency value.

12. The method of claim 9 , wherein localizing the vehicle comprises:
computing a center-of-mass based on the plurality of candidate locations, each candidate location of the plurality of candidate locations weighted by a corresponding consistency value.

13. The method of claim 1 , wherein a distance from the vehicle to the detected traffic boundary in occupancy grid coordinates is based at least in part on an estimate of the camera's pose.

14. The method of claim 1 , wherein the traffic boundary is detected with a neural network that is trained to detect traffic boundaries and objects that tend to appear on or near roads.

15. The method of claim 14 , wherein the visual data is processed by the neural network on an edge computing device that is in the vehicle.

16. The method of claim 1 , where the vehicle is travelling on a road with a plurality of lanes, and wherein localizing the vehicle comprises determining a lane of the plurality of lanes in which the vehicle is travelling.

17. The method of claim 1 , wherein the traffic boundary is a road boundary.

18. The method of claim 1 , further comprising:
updating the portion of the traffic boundary map based on the occupancy grid and the at least one consistency value.

19. A computer program product, the computer program product comprising:
a non-transitory computer-readable medium having program code recorded thereon, the program code, when executed by a processor, causes the processor to:
receive visual data from a camera on a vehicle;
detect a traffic boundary within the visual data;
select a first one or more cells of an occupancy grid based on a bird's eye view projection of the detected traffic boundary, wherein each cell of the occupancy grid corresponds to a position on a road relative to the vehicle;
increment a value of the first one or more cells;
receive a portion of a traffic boundary map;
compute a cross-correlation between the occupancy grid and the portion of the traffic boundary map to produce at least one consistency value; and
localize the vehicle within the traffic boundary map based on the at least one consistency value.

20. An apparatus comprising:
at least one memory unit; and
at least one processor coupled to the at least one memory unit, in which the at least one processor is configured to:
receive visual data from a camera on a vehicle;
detect a traffic boundary within the visual data;
select a first one or more cells of an occupancy grid based on a bird's eye view projection of the detected traffic boundary, wherein each cell of the occupancy grid corresponds to a position on a road relative to the vehicle;
increment a value of the first one or more cells;
receive a portion of a traffic boundary map;
compute a cross-correlation between the occupancy grid and the portion of the traffic boundary map to produce at least one consistency value; and
localize the vehicle within the traffic boundary map based on the at least one consistency value.

Assignees

• Netradyne Inc

Inventors

• Cox Jonathan Albert
• Taranalli Veeresh
• Julian David Jonathan
• Chakravarthy Badugu Naveen
• Campos Michael
• Kahn Adam David
• Venkatachalam Jayaraman Venkata Ramanan
• Yedla Arvind

Classifications

• G01C21/30 Primary

  Map- or contour-matching · CPC title

• G06F18/214

  Generating training patterns; Bootstrap methods, e.g. bagging or boosting · CPC title

• G06V10/764

  using classification, e.g. of video objects · CPC title

• G06V10/774

  Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting · CPC title

• G06V10/80

  Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level (multimodal speaker identification or verification G10L17/10) · CPC title