IP Library › Granted Patent US 12,737,909
Granted Patent B2
US 12,737,909 · App. 18/339,831 · Granted Sep 15, 2026

Systems and methods for using image data to identify lane width

Inventor: Joseph Stamenkovich (Blacksburg, VA)
Assignee: TORC Robotics, Inc.
G06T7/60G01C21/3815G01C21/3837G06V10/774G06V10/82G06V20/588G06V20/70G06T2207/20081G06T2207/20084G06T2207/30256
View Patent ↗
Loading inventors, assignments & file history…
Monitor This Case
Get email alerts when status or documents change.
Order Certified Copies
Most orders are placed with the USPTO same day — all within 24 business hours.
Order via The Patent Place →
Pre-filled with this patent's details
Quick Facts
Patent No.
US 12,737,909
App. No.
18/339,831
Granted
Sep 15, 2026
Kind
B2
Abstract

A method, comprises identifying a set of image data captured by at least one autonomous vehicle when the at least one autonomous vehicle was positioned in a lane of a roadway, and respective ground truth localization data of the at least one autonomous vehicle; determining a plurality of lane width values for the set of image data; labeling the set of image data with the plurality of lane width values, the plurality of lane width values representing a width of a lane in which the at least one autonomous vehicle was positioned; and training using the labeled set of image data, a machine learning model, such that the machine learning model is configured to predict a new lane width value for a new lane as output.

Claims (49)

1 . A method, comprising:

identifying, by one or more processors coupled to non-transitory memory, a set of image data captured by at least one autonomous vehicle when the at least one autonomous vehicle was positioned in a lane of a roadway, and respective ground truth localization data of the at least one autonomous vehicle;

determining, by the one or more processors, a plurality of lane width values for the set of image data, the plurality of lane width values including a first lane width value of the lane in which the at least one autonomous vehicle was positioned and a second lane width value of a lane not being traveled by the at least one autonomous vehicle, determining the plurality of lane width values further comprising:

determining distances between i) lane lines of the lane in which the at least one autonomous vehicle was positioned and ii) the ground truth localization data of the at least one autonomous vehicle, wherein determining the distances further comprises:

generating one or more confidence values of one or more ground truth information sources; and

selecting a ground truth information source of the ground truth localization data from the one or more ground truth information sources, based on the one or more confidence values;

determining the first lane width value, based on the determined distances and a width of the at least one autonomous vehicle; and

determining the second lane width value by comparing distances of lane lines of the lane not being traveled by the at least one autonomous vehicle to a same object within the set of image data;

labeling, by the one or more processors, the set of image data with the plurality of lane width values; and

training, by the one or more processors, using the labeled set of image data, a machine learning model, such that the machine learning model is configured to predict a new lane width value for a new lane as output.

2 . The method of claim 1 , wherein the plurality of lane width values is determined using an image recognition or image segmentation protocol.

3 . The method of claim 1 , wherein the ground truth localization data includes data derived from a high-definition (HD) map.

4 . The method of claim 3 , wherein a plurality of lane indications of the set of image data are defined at least in part as a feature on a raster layer of the high-definition (HD) map.

5 . The method of claim 1 , wherein the machine learning model comprises a plurality of neural network layers.

6 . The method of claim 1 , further comprising:

executing, by the one or more processors, the machine learning model for a second autonomous vehicle.

7 . The method of claim 1 , wherein the one or more ground truth localization sources are separate from the image data.

8 . A non-transitory machine-readable storage medium having computer-executable instructions stored thereon that, when executed by one or more processors, cause the one or more processors to perform operations comprising:

identify a set of image data captured by at least one autonomous vehicle when the at least one autonomous vehicle was positioned in a lane of a roadway, and respective ground truth localization data of the at least one autonomous vehicle;

determine a plurality of lane width values for the set of image data, the plurality of lane width values including a first lane width value of the lane in which the at least one autonomous vehicle was positioned and a second lane width value of a lane not being traveled by the at least one autonomous vehicle, determine the plurality of lane width values further comprising:

determine distances between i) lane lines of the lane in which the at least one autonomous vehicle was positioned and ii) the ground truth localization data of the at least one autonomous vehicle, wherein determine the distances further comprises:

generate one or more confidence values of one or more ground truth information sources; and

select a ground truth information source of the ground truth localization data from the one or more ground truth information sources, based on the one or more confidence values;

determine the first lane width value, based on the determined distances and a width of the at least one autonomous vehicle; and

determine the second lane width value by comparing distances of lane lines of the lane not being traveled by the at least one autonomous vehicle to a same object within the set of image data,

label the set of image data with the plurality of lane width values; and

train, using the labeled set of image data, a machine learning model, such that the machine learning model is configured to predict a new lane width value for a new lane as output.

9 . The non-transitory machine-readable storage medium of claim 8 , wherein the plurality of lane width values is determined using an image recognition or image segmentation protocol.

10 . The non-transitory machine-readable storage medium of claim 8 , wherein the ground truth localization data includes data derived from a high-definition (HD) map.

11 . The non-transitory machine-readable storage medium of claim 10 , wherein a plurality of lane indications of the set of image data are defined at least in part as a feature on a raster layer of the high-definition (HD) map.

12 . The non-transitory machine-readable storage medium of claim 8 , wherein the machine learning model comprises a plurality of neural network layers.

13 . The non-transitory machine-readable storage medium of claim 8 , wherein the instructions further cause the one or more processors to:

executing, by the one or more processors, the machine learning model for a second autonomous vehicle.

14 . The non-transitory machine-readable storage medium of claim 8 , wherein the one or more ground truth localization sources are separate from the image data.

15 . A system comprising a processor configured to:

identify a set of image data captured by at least one autonomous vehicle when the at least one autonomous vehicle was positioned in a lane of a roadway, and respective ground truth localization data of the at least one autonomous vehicle;

determine a plurality of lane width values for the set of image data, the plurality of lane width values including a first lane width value of the lane in which the at least one autonomous vehicle was positioned and a second lane width value of a lane not being traveled by the at least one autonomous vehicle, determine the plurality of lane width values further comprising:

determine distances between i) lane lines of the lane in which the at least one autonomous vehicle was positioned and ii) the ground truth localization data of the at least one autonomous vehicle, wherein determine the distances further comprises:

generate one or more confidence values of one or more ground truth information sources; and

select a ground truth information source of the ground truth localization data from the one or more ground truth information sources, based on the one or more confidence values;

determine the first lane width value, based on the determined distances and a width of the at least one autonomous vehicle; and

determine the second lane width value by comparing distances of lane lines of the lane not being traveled by the at least one autonomous vehicle to a same object within the set of image data,

label the set of image data with the plurality of lane width values; and

train, using the labeled set of image data, a machine learning model, such that the machine learning model is configured to predict a new lane width value for a new lane as output.

16 . The system of claim 15 , wherein the plurality of lane width values is determined using an image recognition or image segmentation protocol.

17 . The system of claim 15 , wherein the ground truth localization data includes data derived from a high-definition (HD) map.

18 . The system of claim 17 , wherein a plurality of lane indications of the set of image data are defined at least in part as a feature on a raster layer of the high-definition (HD) map.

19 . The system of claim 15 , wherein the processor is further configured to: execute the machine learning model for a second autonomous vehicle.

20 . The system of claim 15 , wherein the one or more ground truth localization sources are separate from the image data.

Assignments (1)
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Oct 17, 2024
From: STAMENKOVICH, JOSEPH
To: TORC ROBOTICS, INC.
Reel/Frame 068929/0105 →
Continuity (5)
Continuation In Part 18303460 · Apr 19, 2023
Provisional Application 63447766 · Feb 23, 2023
Provisional Application 63434843 · Dec 22, 2022
Provisional Application 63376860 · Sep 23, 2022
Related Publication 20240104757A1 · Mar 28, 2024
References Cited (45)
US 11403774B2 · Westmacott et al. · 2022 [cited by applicant]
US 11443531B1 · Saggu et al. · 2022 [cited by applicant]
US 11551459B1 · Saggu · 2023 [cited by applicant]
US 11645839B2 · Soni · 2023 [cited by examiner]
US 11738770B2 · Han et al. · 2023 [cited by applicant]
US 20140303828A1 · Joshi · 2014 [cited by applicant]
US 20180189578A1 · Yang et al. · 2018 [cited by applicant]
US 20180307917A1 · Mittal · 2018 [cited by examiner]
US 20200109952A1 · Haque et al. · 2020 [cited by applicant]
US 20200116499A1 · Jung et al. · 2020 [cited by applicant]
US 20200174487A1 · Viswanathan et al. · 2020 [cited by applicant]
US 20200219399A1 · Pfeifle · 2020 [cited by examiner]
US 20200272142A1 · Beauvillain · 2020 [cited by applicant]
US 20200324795A1 · Bojarski · 2020 [cited by examiner]
US 20200341466A1 · Pham · 2020 [cited by examiner]
US 20210009163A1 · Urtasun et al. · 2021 [cited by applicant]
US 20210157330A1 · Tran · 2021 [cited by applicant]
US 20210179097A1 · Pan et al. · 2021 [cited by applicant]
US 20210224556A1 · Xu et al. · 2021 [cited by applicant]
US 20210237737A1 · Al-Nuaimi et al. · 2021 [cited by applicant]
US 20210237768A1 · Tateishi · 2021 [cited by examiner]
US 20210342600A1 · Westmacott et al. · 2021 [cited by applicant]
US 20220153315A1 · Zeng et al. · 2022 [cited by applicant]
US 20220180646A1 · Halfaoui et al. · 2022 [cited by applicant]
US 20220219700A1 · Tsukamoto · 2022 [cited by applicant]
US 20220374428A1 · Nassar et al. · 2022 [cited by applicant]
US 20230099494A1 · Kocamaz et al. · 2023 [cited by applicant]
US 20230101472A1 · Colling et al. · 2023 [cited by applicant]
US 20230122011A1 · Ishimaru et al. · 2023 [cited by applicant]
US 20230154206A1 · Kim et al. · 2023 [cited by applicant]
US 20230196913A1 · Shnaider · 2023 [cited by applicant]
US 20230298363A1 · Zhang · 2023 [cited by examiner]
US 20230347887A1 · Favreau et al. · 2023 [cited by applicant]
US 20230391358A1 · Donderici · 2023 [cited by applicant]
US 20240096109A1 · Sharma · 2024 [cited by examiner]
CN 114898310A · 2022 [cited by applicant]
JP 2017059003A · 2017 [cited by examiner]
JP 2019194751A · 2019 [cited by applicant]
WO 2021006870A1 · 2021 [cited by applicant]
WO 2022211927A1 · 2022 [cited by applicant]
Soomok Lee, Ego-lane index-aware vehicular localisation using the DeepRoad Network for urban environments, Jan. 4, 2021, Wiley, The Institution of Engineering and Technology (Year: 2021). [cited by applicant]
Baoguo Yu, Ego-Lane Index Estimation Based on Lane-Level Map and LiDAR Road Boundary Detection, Oct. 27, 2021, M DPI, Sensors (Year: 2021). [cited by applicant]
Borkar, A Novel Lane Detection System With Efficient Ground Truth Generation, Dec. 8, 2011, IEEE, IEEE Transactions on Intelligent Transportation Systems vol. 13 No. 1 (Year: 2011). [cited by applicant]
Seo et al., “Tracking and estimation of ego-vehicle's state for lateral localization,” 17th International IEEE Conference on Intelligent Transportation Systems (ITSC), Qingdao, China, 2014, pp. 1251-1257, doi: 10.1109/I… [cited by applicant]
Fujimoto et al., “LaneFusion: 3D Object Detection with Rasterized Lane Map,” 2022 IEEE Intelligent Vehicles Symposium (IV), Aachen, Germany, 2022, pp. 396-403, doi: 10.1109/IV51971.2022.9827025. (Year: 2022). [cited by applicant]