Methods and systems for camera-based autonomous parking

1. A vehicle comprising:

a camera;

a range sensor;

a GPS receiver;

one or more actuator systems; and

one or more processors operatively coupled to the camera, the range sensor, the GPS receiver, and the one or more actuator systems, the one or more processors configured to perform a method comprising:

detecting two or more parking lines in one or more images captured by the camera;

localizing the vehicle with respect to the two or more parking lines based on:

location data of the vehicle determined from the GPS receiver, and

a location determination for the vehicle based on detected ends of the two or more parking lines;

determining an occupancy state of one or more parking spaces formed by the two or more parking lines using the range sensor;

in accordance with a determination that the occupancy state of a respective parking space of the one or more parking spaces is empty, selecting the respective parking space;

identifying a region of interest including the selected parking space;

detecting one or more selected parking lines of the selected parking space within one or more captured images including all or part of the region of interest;

calculating one or more errors of a current location of the vehicle based on the one or more selected parking lines; and

moving the vehicle, using the one or more actuator systems, in a direction to reduce the one or more errors.

2. The vehicle of claim 1 , wherein detecting the two or more parking lines comprises:

determining pixel gradients in the one or more images; and

detecting the two or more parking lines based on the determined pixel gradients.

3. The vehicle of claim 1 , wherein detecting the ends of the two or more parking lines comprises:

matching top and bottom edge pixels for each lane line;

fitting linear models to the top and bottom edge pixels; and

identifying an end of each lane line based on a pixel gradient template.

4. The vehicle of claim 1 , wherein determining the location for the vehicle based on the detected ends of the two or more parking lines comprises:

initializing a plurality of candidate vehicle positions; and

assigning probabilities to the plurality of candidate vehicle positions based on the detected ends of the two or more parking lines.

5. The vehicle of claim 1 , wherein determining the location for the vehicle based on the location data for the vehicle from the GPS receiver comprises:

initializing a plurality of candidate vehicle positions; and

assigning probabilities to the plurality of candidate vehicle positions based on a GPS location of the vehicle.

6. The vehicle of claim 1 , wherein determining the occupancy state of the one or more parking spaces formed by the two or more parking lines using the range sensor comprises:

initializing occupancy grid cells of an occupancy grid;

updating the occupancy grid cells of the occupancy grid based on detections made by the range sensor; and

determining the occupancy state of the one or more parking spaces based on the occupancy grid cells of the occupancy grid.

7. The vehicle of claim 1 , wherein identifying the region of interest comprises:

estimating a location of the region of interest based on one or more of the captured images that were captured before selecting the selected parking space; and

after selecting the selecting parking space, capturing one or more of the captured images including all or part of the region of interest.

8. The vehicle of claim 1 , wherein the one or more errors comprise one or more of:

a lateral error comprising a first distance between a rear axle of the vehicle and a first parking line of the selected parking space and a second distance between the rear axle and a horizontal parking line of the selected parking space, the first and second lines of the selected parking space being oriented along a first axis;

a longitudinal error comprising a third distance between the rear axle of the vehicle and a third parking line of the selected parking space, the third parking line oriented along a second axis orthogonal to the first axis; and

a heading error comprising an angle between the vehicle and one of the first axis or second axis.

9. The vehicle of claim 1 , wherein detecting the one or more parking lines of the selected parking space comprises applying one or more constraints based on an estimated region of interest to one or more images captured after selecting the selected parking space.

10. The vehicle of claim 1 , wherein the method performed by the one or more processors further comprises:

when the one or more errors are below a predetermined threshold, transition the vehicle to a parked state using the one or more actuator systems.

11. A non-transitory computer-readable storage medium storing instructions that, when executed by one or more processors included in a vehicle, causes the one or more processors to perform a method, the method comprising:

detecting two or more parking lines in one or more images captured by a camera;

localizing a vehicle with respect to the two or more parking lines based on:

location data for the vehicle from a GPS receiver, and

a location determination for the vehicle based on detected ends of the two or more parking lines;

determining an occupancy state of one or more parking spaces formed by the two or more parking lines using a range sensor; and

in accordance with a determination that the occupancy state of a respective parking space of the one or more parking spaces is empty, selecting the respective parking space;

identifying a region of interest including the selected parking space;

detecting one or more selected parking lines of the selected parking space within one or more captured images including all or part of the region of interest;

calculating one or more errors of a current location of the vehicle based on the one or more selected parking lines; and

moving the vehicle, using one or more actuator systems of the vehicle, in a direction to reduce the one or more errors.

12. The non-transitory computer-readable storage medium of claim 11 , wherein detecting the two or more parking lines comprises:

determining pixel gradients in the one or more images; and

detecting the two or more parking lines based on the determined pixel gradients.

13. The non-transitory computer-readable storage medium of claim 11 , wherein detecting the ends of the two or more parking lines comprises:

matching top and bottom edge pixels for each lane line;

fitting linear models to the top and bottom edge pixels; and

identifying an end of each lane line based on a pixel gradient template.

14. The non-transitory computer-readable storage medium of claim 11 , wherein determining the location for the vehicle based on the detected ends of the two or more parking lines comprises:

initializing a plurality of candidate vehicle positions; and

assigning probabilities to the plurality of candidate vehicle positions based on the detected ends of the two or more parking lines.

15. The non-transitory computer-readable storage medium of claim 11 , wherein determining the location for the vehicle based on the location data for the vehicle from the GPS receiver comprises:

initializing a plurality of candidate vehicle positions; and

assigning probabilities to the plurality of candidate vehicle positions based on a GPS location of the vehicle.

16. The non-transitory computer-readable storage medium of claim 11 , wherein determining the occupancy state of the one or more parking spaces formed by the two or more parking lines using the range sensor comprises:

initializing occupancy grid cells of an occupancy grid;

updating the occupancy grid cells of the occupancy grid based on detections made by the range sensor; and

determining the occupancy state of the one or more parking spaces based on the occupancy grid cells of the occupancy grid.

17. The non-transitory computer-readable storage medium of claim 1 , wherein identifying the region of interest comprises:

estimating a location of the region of interest based on one or more of the captured images that were captured before selecting the selected parking space; and

after selecting the selecting parking space, capturing one or more of the captured images including all or part of the region of interest.

18. The non-transitory computer-readable storage medium of claim 11 , wherein the one or more errors comprise one or more of:

a lateral error comprising a first distance between a rear axle of the vehicle and a first parking line of the selected parking space and a second distance between the rear axle and a horizontal parking line of the selected parking space, the first and second lines of the selected parking space being oriented along a first axis;

a longitudinal error comprising a third distance between the rear axle of the vehicle and a third parking line of the selected parking space, the third parking line oriented along a second axis orthogonal to the first axis; and

a heading error comprising an angle between the vehicle and one of the first axis or second axis.

19. The non-transitory computer-readable storage medium of claim 11 , wherein detecting the one or more parking lines of the selected parking space comprises applying one or more constraints based on an estimated region of interest to one or more images captured after selecting the selected parking space.

20. The non-transitory computer-readable storage medium of claim 11 , wherein the method performed by the one or more processors further comprises:

when the one or more errors are below a predetermined threshold, transition the vehicle to a parked state using the one or more actuator systems.