Farmland Vehicle Heading Angle Estimation From Monocular Road Images

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing methods for calculating vehicle heading angles in farmland environments using monocular cameras suffer from poor depth estimation accuracy due to reliance on three-dimensional reconstruction or insufficient training sample sizes, leading to inaccurate navigation.

Innovation Solution

A method that involves acquiring a current road surface image, determining multiple processed road surface points, and calculating depth estimation values based on the camera's position state, height, and image parameters to improve depth estimation accuracy.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Loss of information

If three-dimensional reconstruction is used for depth estimation, then depth information can be obtained, but depth estimation accuracy deteriorates when capturing images along the visual axis in farmland environments

Engineering Contradiction:
Improvedepth informationVSAvoiddepth estimation accuracy
Core Design Contradiction:
Loss of informationVSMeasurement precision

Solution Approach 1:

The patent changes the parameter of image capture direction from along the visual axis to oblique angles relative to the visual axis. By capturing images at oblique angles, the system obtains better depth estimation results through three-dimensional reconstruction, resolving the contradiction between obtaining depth information and maintaining depth estimation accuracy in farmland environments.

Inventive Principle:
Principle #35Parameter changes

2Loss of information

If deep learning methods are used for depth estimation, then depth information can be obtained, but depth estimation accuracy deteriorates when training sample size is insufficient

Engineering Contradiction:
Improvedepth informationVSAvoiddepth estimation accuracy
Core Design Contradiction:
Loss of informationVSMeasurement precision

Solution Approach 1:

The patent introduces an intermediary approach by combining oblique angle image capture with three-dimensional reconstruction methodology. This intermediary solution bridges the gap between monocular camera limitations and accurate depth estimation, achieving reliable depth information without relying on large-scale training datasets required by deep learning methods.

Inventive Principle:
Principle #24Intermediary (Mediator)

3Device complexity

If monocular camera is used for navigation, then device complexity is reduced, but navigation accuracy deteriorates due to poor depth estimation

Engineering Contradiction:
Improvecamera system complexityVSAvoidnavigation accuracy
Core Design Contradiction:
Device complexityVSMeasurement precision

Solution Approach 1:

The patent changes the capture angle parameter from visual axis alignment to oblique angles, which fundamentally improves depth estimation quality while maintaining monocular camera simplicity. This parameter change enables accurate navigation using only a single camera, resolving the contradiction between device simplicity and navigation accuracy.

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentEP4697265A1Method for calculating vehicle heading angle in road surface environment of farmland
Publication Date: 2026.02.18 SHANGHAI HUACE NAVIGATION TECH
  • EP4697265A1 patent drawingFigure 1A~1B
  • EP4697265A1 patent drawingFigure 1C~1D
  • EP4697265A1 patent drawingFigure 1E~1F

AI summary

Provided is a method for calculating a vehicle heading angle in a road surface environment of farmland. The method includes acquiring (110, 210) a current road surface image in front of a to-be-controlled vehicle by using a camera disposed on the to-be-controlled vehicle, and determining multiple currently processed road surface points according to the current road surface image; acquiring (120, 220) the position state of each currently processed road surface point, a first height value of the camera from the ground, a preset parameter of the camera, and description information of the each currently processed road surface point in the current road surface image; determining (130, 230) the depth estimation value of the each currently processed road surface point according to the position state, the first height value of the camera from the ground, the preset parameter, and the description information, and determining a current heading angle matching the each currently processed road surface point according to the depth estimation value; and determining (140, 260) a target heading angle matching the current road surface image according to multiple current heading angles corresponding to the multiple currently processed road surface points.