Trailer detection and autonomous hitching

1. A method for autonomously maneuvering a tow vehicle towards a trailer for autonomous hitching between the trailer and the tow vehicle, the method comprising:

receiving, at a data processing hardware, images from one or more cameras positioned on a back portion of the tow vehicle and in communication with the data processing hardware;

receiving, at the data processing hardware, sensor data from an inertial measurement unit in communication with the data processing hardware and supported by the tow vehicle;

determining, at the data processing hardware, a pixel-wise intensity difference between a current received image and a previous received image;

determining, at the data processing hardware, a camera pose and a trailer pose with respect to a world coordinate system, the camera pose and the trailer pose based on the images, the sensor data, and the pixel-wise intensity difference;

determining, at the data processing hardware, a tow vehicle path based on the camera pose and the trailer pose;

sending, from the data processing hardware to a drive system supported by the tow vehicle, instruction to autonomously maneuver the tow vehicle along the tow vehicle path in a reverse direction causing the tow vehicle to hitch with the trailer;

wherein the camera pose includes a camera location and a camera angle in the world coordinate system and the trailer pose includes a trailer location and a trailer angle in the world coordinate system;

identifying one or more feature points within the images, the feature points associated with the trailer;

determining a plane based on the one or more feature points, the plane being indicative of and corresponding to a front face of the trailer at a distance from a trailer coupler of the trailer;

determining a normal line to the plane; and

determining the trailer angle based on the normal line.

2. The method of claim 1 , wherein the one or more cameras include a fisheye lens that includes an ultra wide-angle lens.

3. The method of claim 1 , further comprising:

determining a region of interest surrounding a representation of the trailer within the images,

wherein determining a pixel-wise intensity difference between the current received image and a previous received image, includes determining the pixel-wise intensity difference between the region of interest of the current received image and the region of interest of the previous received image.

4. The method of claim 3 , further comprising:

tracking the one or more feature points as the vehicle autonomously maneuvers along the path.

5. The method of claim 1 , further comprising:

determining a sequence of points associated with the path, wherein the tow vehicle follows the sequence of points;

determining a steering wheel angle associated with each point within the sequence of points; and

sending instructions to the drive system to adjust a current steering wheel angle based on the determined steering wheel angle.

6. The method of claim 1 , further comprising:

determining a difference between a steering angle of each front wheel of the tow vehicle, as the vehicle autonomously maneuver along the path; and

sending instructions to the drive system to adjust the steering wheel angle of each wheel based on the difference between the steering wheel angle of each front wheel.

7. The method of claim 1 , wherein determining the tow vehicle path comprises a joining of a plurality of circular arcs of maximum curvature and one or more line segments.

8. A system for autonomously maneuvering a tow vehicle towards a trailer for autonomous hitching between the trailer and the tow vehicle, the system comprising:

data processing hardware; and

memory hardware in communication with the data processing hardware, the memory hardware storing instructions that when executed on the data processing hardware cause the data processing hardware to perform operations comprising:

receiving images from one or more cameras positioned on a back portion of the tow vehicle and in communication with the data processing hardware;

receiving sensor data from an inertial measurement unit in communication with the data processing hardware and supported by the tow vehicle;

determining a pixel-wise intensity difference between a current received image and a previous received image;

determining a camera pose and a trailer pose with respect to a world coordinate system, the camera pose and the trailer pose based on the images, the sensor data, and the pixel-wise intensity difference;

determining a tow vehicle path based on the camera pose and the trailer pose;

sending instruction to a drive system supported by the tow vehicle, the instructions causing the tow vehicle to autonomously maneuver along the tow vehicle path in a reverse direction causing the tow vehicle to hitch with the trailer;

determining a difference between a steering angle of each front wheel of the tow vehicle, as the vehicle autonomously maneuver along the path; and

sending instructions to the drive system to adjust the steering wheel angle of each wheel based on the difference between the steering wheel angle of each front wheel.

9. The system of claim 8 , wherein the one or more cameras include a fisheye lens that includes an ultra wide-angle lens.

10. The system of claim 8 , wherein:

the camera pose includes a camera location and a camera angle in the world coordinate system; and

the trailer pose includes a trailer location and a trailer angle in the world coordinate system.

11. The system of claim 10 , wherein the operations further include:

identifying one or more feature points within the images, the feature points associated with the trailer;

determining a plane based on the one or more feature points, the plane indicative of and corresponding to a front face of the trailer that is at a distance from a trailer coupler of the trailer;

determining a normal line to the plane; and

determining the trailer angle based on the normal line.

12. The system of claim 8 , wherein the operations further include:

determining a region of interest surrounding a representation of the trailer within the images,

wherein determining a pixel-wise intensity difference between the current received image and a previous received image, includes determining the pixel-wise intensity difference between the region of interest of the current received image and the region of interest of the previous received image.

13. The system of claim 12 , wherein the operations further include:

identifying one or more feature points within the region of interest, the feature points associated with the trailer;

tracking the one or more feature points as the vehicle autonomously maneuvers along the path.

14. The system of claim 8 , wherein the operations further include:

determining a sequence of points associated with the path, wherein the tow vehicle follows the sequence of points;

determining a steering wheel angle associated with each point within the sequence of points; and

sending instructions to the drive system to adjust a current steering wheel angle based on the determined steering wheel angle.