Systems and methods for electric vehicle charging using machine learning

1. A method for a charging system to autonomously charge an electric vehicle (EV), comprising:

obtaining, by a control system, a trained machine learning (ML) model from a back-end server;

capturing, by the control system, an image using an image capturing device of the charging system, wherein a portion of the image comprises an EV charging portal of the EV;

inputting, by the control system, the image into the trained ML model to determine one or more regions of interest associated with the EV charging portal within the image;

determining, by the control system, a location of the EV charging portal based on the one or more determined regions of interest and one or more image processing techniques; and

providing, by the control system, information to maneuver a robotic arm of the charging system to a physical position based on the determined location of the EV charging portal.

2. The method of claim 1 , wherein inputting the image into the trained ML model to determine the one or more regions of interest comprises inputting the image into the trained ML model to generate one or more dimension dependent parameters associated with one or more bounding boxes within the image, and

wherein determining the location of the EV charging portal is based on:

filtering out pixels within the image that are outside of the one or more bounding boxes; and

applying the one or more image processing techniques to remaining pixels within the one or more bounding boxes.

3. The method of claim 2 , wherein the one or more dimension dependent parameters indicate an area of a first bounding box of the one or more bounding boxes, one or more dimensions of the first bounding box, or one or more corner points of the first bounding box.

4. The method of claim 1 , wherein inputting the image into the trained ML model to determine the one or more regions of interest comprises inputting the image into the trained ML model to generate a pixel value dependent parameter associated with a pixel characteristic, and

wherein determining the location of the EV charging portal is based on:

filtering out pixels within the image based on the pixel value dependent parameter; and

applying the one or more image processing techniques to the image based on filtering out the pixels.

5. The method of claim 4 , wherein the pixel characteristic is a value associated with a pixel within the image, and wherein the pixel value dependent parameter is a datatype threshold that is used to filter out the pixels within the image.

6. The method of claim 1 , wherein the robotic arm is controllably extendable and retractable, wherein the robotic arm is adapted to carry a charging plug located at a distal end of the robotic arm, wherein the charging plug is configured to be controllably moveable and insertable into the EV charging portal.

7. The method of claim 1 , wherein capturing the image using the image capturing device comprises:

providing instructions for a light emitter to emit illumination at a light spectrum;

capturing the image using the illumination at the light spectrum; and

using a band-pass filter to filter out portions of the image based on the illumination at the light spectrum,

wherein inputting the image into the trained ML model comprises inputting the image that has been filtered using the band-pass filter into the trained ML model.

8. The method of claim 1 , wherein the trained ML model is a convolutional neural network (CNN), and wherein the back-end server trains the CNN using actual images of one or more first EV charging portals and/or synthetic data of one or more second EV charging portals.

9. The method of claim 1 , wherein determining the location of the EV charging portal comprises:

determining a plurality of keypoints within the determined regions of interest based on a known geometry associated with the EV charging portal;

generating a virtual skeleton image of the EV charging portal based on the plurality of keypoints; and

determining the location of the EV charging portal based on applying the one or more image processing techniques to the plurality of keypoints and the virtual skeleton image.

10. The method of claim 9 , wherein the one or more image processing techniques comprises a scale-invariant feature transform (SIFT) technique or an oriented FAST and rotated BRIEF (ORB) technique.

11. The method of claim 1 , wherein determining the location of the EV charging portal comprises:

determining a six degrees of freedom (DoF) associated with the EV charging portal based on the one or more determined regions of interest and one or more image processing techniques; and

determining the location of the EV charging portal based on the six DoF.

12. The method of claim 1 , wherein inputting the image into the trained ML model to determine the one or more regions of interest comprises inputting the image into the trained ML model to generate one or more dimension dependent parameters associated with one or more segmentation masks within the image, and

wherein determining the location of the EV charging portal is based on:

filtering out pixels within the image that are outside of the one or more segmentation masks; and

applying the one or more image processing techniques to remaining pixels within the one or more segmentation masks.

13. The method of claim 12 , wherein the one or more dimension dependent parameters indicate an area of a first segmentation mask of the one or more segmentation masks, one or more dimensions of the first segmentation mask, or one or more corner points of the first segmentation mask.

14. The method of claim 1 , wherein inputting the image into the trained ML model to determine the one or more regions of interest comprises inputting the image into the trained ML model to generate one or more dimension dependent parameters associated with one or more point clouds within the image, and

wherein determining the location of the EV charging portal is based on:

filtering out pixels within the image that are outside of the one or more point clouds; and

applying the one or more image processing techniques to remaining pixels within the one or more point clouds.

15. The method of claim 14 , wherein the one or more dimension dependent parameters indicate an area of a first point cloud of the one or more point clouds, one or more dimensions of the first point cloud, or one or more corner points of the first point cloud.

16. A plug connection system for use in autonomously charging an electric vehicle (EV), the plug connection system comprising:

a robotic arm that is controllably extendable and retractable, wherein the robotic arm is adapted to carry a charging plug located at a distal end of the robotic arm, wherein the charging plug is configured to be controllably moveable and insertable into a EV charging portal of the EV;

an image capturing device configured to capture one or more images; and

a control system configured to:

obtain a trained machine learning (ML) model from a back-end server;

capture an image using the image capturing device, wherein a portion of the image comprises the EV charging portal;

input the image into the trained ML model to determine one or more regions of interest associated with the EV charging portal within the image;

determine a location of the EV charging portal based on the one or more determined regions of interest and one or more image processing techniques; and

provide information to maneuver the robotic arm to a physical position based on the determined location of the EV charging portal.

17. The plug connection system of claim 16 , wherein inputting the image into the trained ML model to determine the one or more regions of interest comprises inputting the image into the trained ML model to generate one or more dimension dependent parameters associated with one or more bounding boxes within the image, and

wherein determining the location of the EV charging portal is based on:

filtering out pixels within the image that are outside of the one or more bounding boxes; and

applying the one or more image processing techniques to remaining pixels within the one or more bounding boxes.

18. The plug connection system of claim 17 , wherein the one or more dimension dependent parameters indicate an area of a first bounding box of the one or more bounding boxes, one or more dimensions of the first bounding box, or one or more corner points of the first bounding box.

19. The plug connection system of claim 16 , wherein inputting the image into the trained ML model to determine the one or more regions of interest comprises inputting the image into the trained ML model to generate a pixel value dependent parameter associated with a pixel characteristic, and

wherein determining the location of the EV charging portal is based on:

filtering out pixels within the image based on the pixel value dependent parameter; and

applying the one or more image processing techniques to the image based on filtering out the pixels.

20. A non-transitory computer-readable medium having processor-executable instructions stored thereon, wherein the processor-executable instructions, when executed by one or more controllers, facilitate:

obtaining a trained machine learning (ML) model from a back-end server;

capturing an image using an image capturing device of a charging system, wherein a portion of the image comprises the EV charging portal;

inputting the image into the trained ML model to determine one or more regions of interest associated with the EV charging portal within the image;

determining a location of the EV charging portal based on the one or more determined regions of interest and one or more image processing techniques; and

providing information to maneuver a robotic arm of the charging system to a physical position based on the determined location of the EV charging portal.