Trailer angle detection using end-to-end learning

Patent metadata

Field	Value
Publication number	US-2020282910-A1
Application number	US-201916294540-A
Country	US
Kind code	A1
Filing date	Mar 6, 2019
Priority date	Mar 6, 2019
Publication date	Sep 10, 2020
Grant date	—

Abstract

A method for training an image-based trailer identification system comprises capturing a plurality of captured images in a field of view and identifying a detected trailer angle for a trailer in connection with a vehicle in each of the captured images. The method further comprises comparing the captured images and the corresponding trailer angles to a predetermined image set comprising a plurality of teaching trailer angles and identifying at least one required trailer angle of the teaching trailer angles that is not included in the captured images. Based on the captured images, a simulated angle image is generated. The simulated image comprises a depiction of the trailer in connection with the vehicle at the at least one required angle not included in the captured images. The method further comprises supplying the simulated angle image to the identification system for training.

First claim

The invention claimed is:

1 . A method for training an image-based trailer identification system comprising:
capturing a plurality of captured images in a field of view; identifying a detected trailer angle for a trailer in connection with a vehicle in each of the captured images; comparing the captured images and the corresponding trailer angles to a predetermined image set comprising a plurality of teaching trailer angles; identifying at least one required trailer angle of the teaching trailer angles that is not included in the captured images; generating a simulated angle image based on the captured images, wherein the simulated image comprises a depiction of the trailer in connection with the vehicle at the at least one required angle not included in the captured images; and supplying the simulated angle image to the identification system for training.

2 . The method according to claim 1 , wherein the identifying of the detected trailer angle is determined by digitally processing the image data via a feature extraction technique.

3 . The method according to claim 2 , wherein the feature extraction technique comprises a least one of a line detection.

4 . The method according to claim 1 , wherein the identification system comprises a neural network.

5 . The method according to claim 1 , wherein the plurality of teaching trailer angles comprises a range of trailer angles for identification.

6 . The method according to claim 5 , wherein the plurality of teaching trailer angles comprises an increment between each trailer angle in the range of trailer angles.

7 . The method according to claim 1 , further comprising:
detecting a lighting condition in the captured images and comparing the lighting condition to a required lighting condition for training the identification system.

8 . The method according to claim 7 , further comprising:
in response to the lighting condition of the captured images differing from the required lighting condition, adjusting the lighting at least one of the captured images to generate the simulated lighting image.

9 . The method according to claim 8 , further comprising:
supplying the simulated lighting image to the identification system for training.

10 . The method according to claim 1 , wherein the image data is captured by a plurality of vehicles, each configured to capture the images in the field of view.

11 . The method according to claim 10 , further comprising:
storing the captured images from the plurality of vehicles and compiling the captured images for comparing the captured images.

12 . A system for training a neural network comprising:
at least one imaging device configured to capture a plurality of captured images of a trailer in connection with a vehicle; an image processing apparatus configured to: identify a detected trailer angle from each of the captured images based on an image processing routine; compare the captured images and the corresponding detected trailer angles to a predetermined image set comprising a plurality of teaching trailer angles; identify at least one required trailer angle of the teaching trailer angles that is not included in the captured images; generate a simulated angle image based on the captured images; and supply the simulated angle image to the neural network for training.

13 . The system according to claim 12 , wherein the simulated angle image comprises a depiction of the trailer in connection with the vehicle at the at least one required angle not included in the captured images.

14 . The system according to claim 12 , wherein the image processing routine comprises a digital image processing technique comprising a feature extraction technique.

15 . The system according to claim 14 , wherein the feature extraction technique comprises a least one of a line detection.

16 . The system according to claim 12 , wherein the plurality of teaching trailer angles comprises a range of trailer angles for identification.

17 . The system according to claim 16 , wherein the plurality of teaching trailer angles comprises an increment between each trailer angle in the range of trailer angles.

18 . The system according to claim 12 , wherein the image processing apparatus is further configured to:
detect a lighting condition in the captured images and compare the lighting condition to a required lighting condition for training the identification system.

19 . The system according to claim 18 , wherein the image processing apparatus is further configured to:
in response to the lighting condition of the captured images differing from the required lighting condition, adjust the lighting to at least one of the captured images and generate the simulated lighting image.

20 . A system for training a neural network comprising:
at least one imaging device configured to capture a plurality of captured images of a trailer in connection with a coupler of a vehicle; an image processing apparatus configured to:
identify a first trailer type in the captured images based on a feature extraction technique configured to identify at least one feature of the trailer relative to the coupler;
compare the first trailer type to a plurality of teaching trailer types;
generate a simulated trailer image comprising a second trailer type of the plurality of teaching trailer types that is not included in the captured images; and
supply the simulated trailer image to the neural network for training.

Assignees

• Ford Global Tech Llc

Inventors

• Nagasamy Vijay

Classifications

• B60R1/003 Primary

  for viewing trailer hitches · CPC title

• G06N3/045

  Combinations of networks · CPC title

• G06N3/0464

  Convolutional networks [CNN, ConvNet] · CPC title

• G06N3/09

  Supervised learning · CPC title

• G06N3/0895

  Weakly supervised learning, e.g. semi-supervised or self-supervised learning · CPC title