Method and device for determining a vehicle position
The present disclosure describes a method and an apparatus for determining a corrected position of a vehicle based on a stable landmark. The method includes determining a last known position vector of the vehicle; capturing an image within a vicinity of a vehicle using an imaging device; identifying a stable landmark within the captured image based on a previously constructed reference map of the vicinity of the vehicle; determining a correction for a position of the vehicle based on the determined last known position vector of the vehicle and the identified stable landmark; and determining an updated position of the vehicle based on the determined correction.
1. A method for determining a position of a vehicle, comprising:
determining a last known position vector of the vehicle;
capturing an image within a vicinity of the vehicle using an imaging device;
identifying a stable landmark within the captured image based on a previously constructed reference map of the vicinity of the vehicle, wherein the previously constructed reference map is based on georeferenced spatial information captured over a period of time, wherein the stable landmarks are elements that do not change over the period of time;
determining a correction for a position of the vehicle based on the last known position vector and the stable landmark, wherein determining the correction for the position of the vehicle further comprises:
matching the captured image with an image within a visual index that is most similar to the captured image, wherein matching the captured image is based on feature matching and minimizing reprojection error
wherein the reprojection error is represented by:
∥π((ω c −1 ∘ω q −1 ∘ω r )· X i )− x i ∥ 2 ,
where π denotes a projection of the imaging device, ω c denotes a position of the imaging device relative to the vehicle, ω q is a position of a query image, ω r is a position of a reference image stored in the reference map, X i is the last known position vector, and x i is the correction; and
determining an updated position of the vehicle based on the determined correction.
2. The method of claim 1 , wherein determining the correction for a position of the vehicle is performed when a position accuracy of a location determination of the vehicle based on a navigation satellite system (GNSS) falls below a required threshold.
3. The method of claim 1 , wherein the last known position vector of the vehicle is determined at least in part based on an inertial measurement unit (IMU).
4. The method of claim 3 , wherein determining the correction for a position of the vehicle is performed when an accuracy of data retrieved from the IMU falls below a threshold.
5. The method of claim 1 , wherein the imaging device includes at least one of a camera and a LIDAR sensor.
6. An apparatus comprising:
a global navigation satellite system (GNSS) receiver configured to determine a position of the apparatus;
an imaging device;
a processor; and
a memory to store computer program instructions for determining a position of a vehicle, the computer program instructions when executed on the processor cause the processor to perform operations comprising:
determining a last known position vector of the vehicle based on the GNSS receiver;
capturing an image within a vicinity of a vehicle using the imaging device;
identifying a stable landmark within the captured image based on a previously constructed reference map of the vicinity of the vehicle, wherein the previously constructed reference map is based on georeferenced spatial information captured over a period of time, wherein the stable landmarks are elements that do not change over the period of time;
determining a correction for a position of the vehicle based on the determined last known position vector of the vehicle and the identified stable landmark, wherein determining the correction for a position of the vehicle further comprises:
matching the captured image with an image within a visual index that is most similar to the captured image, wherein matching the captured image is based on feature matching and minimizing reprojection error
wherein the reprojection error is represented by the following:
∥π((ω c −1 ∘ω q −1 ∘ω r )· X i )− x i ∥ 2 ,
where π denotes a projection of the imaging device, ω c denotes a position of the imaging device relative to the vehicle, ω q is a position of a query image, ω r is a position of a reference image stored in the reference map, X i is the last known position vector, and x i is the correction; and
determining an updated position of the vehicle based on the determined correction.
7. The apparatus of claim 6 , wherein determining the correction for a position of the vehicle is performed when a position accuracy of a location determination of the vehicle based on a navigation satellite system (GNSS) falls below a required threshold.
8. The apparatus of claim 6 , wherein the last known position vector of the vehicle is determined at least in part based on an inertial measurement unit (IMU).
9. The apparatus of claim 8 , wherein determining the correction for a position of the vehicle is performed when an accuracy of data retrieved from the IMU falls below a threshold.
10. The apparatus of claim 6 , wherein the imaging device includes at least one of a camera and a LIDAR sensor.
11. A non-transitory computer readable medium storing computer program instructions for determining a position of a vehicle, which, when executed on a processor, cause the processor to perform operations comprising:
determining a last known position vector of the vehicle;
capturing an image within a vicinity of a vehicle using an imaging device;
identifying a stable landmark within the captured image based on a previously constructed reference map of the vicinity of the vehicle, wherein the previously constructed reference map is based on georeferenced spatial information captured over a period of time, wherein the stable landmarks are elements that do not change over the period of time;
determining a correction for a position of the vehicle based on the last known position vector and the stable landmark, wherein determining the correction for a position of the vehicle further comprises:
matching the captured image with an image within a visual index that is most similar to the captured image, wherein matching the captured image is based on feature matching and minimizing reprojection error
wherein the reprojection error is represented by the following:
∥π((ω c −1 ∘ω q −1 ∘ω r )· X i )− x i ∥ 2 ,
where π denotes a projection of the imaging device, ω c denotes a position of the imaging device relative to the vehicle, ω q is a position of a query image, ω r is a position of a reference image stored in the reference map, X i is the last known position vector, and x i is the correction; and
determining an updated position of the vehicle based on the determined correction.
12. The non-transitory computer readable medium of claim 11 , wherein determining the correction for a position of the vehicle is performed when a position accuracy of a location determination of the vehicle based on a navigation satellite system (GNSS) falls below a required threshold.
13. The non-transitory computer readable medium of claim 11 , wherein the last known position vector of the vehicle is determined at least in part based on an inertial measurement unit (IMU).
14. The non-transitory computer readable medium of claim 13 , wherein determining the correction for a position of the vehicle is performed when an accuracy of data retrieved from the IMU falls below a threshold.
15. The non-transitory computer readable medium of claim 11 , wherein the imaging device includes at least one of a camera and a LIDAR sensor.