Agricultural Vehicle Steering via Texture Probability Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for automating the steering of agricultural vehicles in fields, such as those using laser scanners and photogrammetry, face challenges in accurately detecting and following tracks with varying visual textures, leading to potential loss of information and reduced reliability, especially when signal strength is weak.
Innovation Solution
A method that analyzes image texture information to assign probability-values to areas of the image, assuming geometric properties of specific structures, and establishes a most probable position parameter, thereby avoiding information loss and improving detection accuracy, even in weak signal conditions. This method includes a learning step to gather texture information and uses additional navigation sensors for steering decisions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Difficulty of detecting and measuring
If texture analysis is used to detect structures in the field, then detection capability is improved, but information loss occurs due to binarisation
Solution Approach 1:
The patent applies preliminary action by performing a learning step before actual detection. During this learning phase, the system gathers texture information from training images and stores it for later use. This preliminary gathering of information allows the system to make more informed detection decisions without losing critical texture details during the actual detection process.
Solution Approach 2:
The patent introduces probability values as an intermediary between raw texture analysis and binary detection results. Instead of directly binarizing texture information, the system assigns probability values that represent the likelihood of structure presence. This intermediary representation preserves more information while still enabling binary decision-making for final detection.
2Measurement precision
If probability-values are assigned to image areas, then detection accuracy is improved, but signal reliability decreases in weak signal conditions
Solution Approach 1:
The patent implements feedback by using the learned texture information to continuously refine probability value assignments. The system compares actual image textures against learned patterns and adjusts probability assignments based on this feedback. This iterative feedback mechanism maintains detection accuracy even when signal conditions are weak or ambiguous.
3Measurement precision
If geometric properties are assumed for structures, then position parameter estimation is improved, but adaptability to varying structures decreases
Solution Approach 1:
The patent applies parameter changes by allowing the geometric property assumptions to be adjusted based on the specific detection task and structure type. The system can modify parameters such as expected structure width, orientation constraints, and shape characteristics to adapt to different agricultural structures while maintaining precise position estimation through the learned texture patterns.
Data Source
Figure 1~2
Figure 1b~2a
Figure 3~4
AI summary
An agricultural vehicle (2) comprises a steering system providing steering signals, said steering system comprising an imaging device (11) for imaging surroundings of the vehicle and an image processing device (13), said steering system operating to provide by means of the imaging device (11) an image of the field (21), analyse the image to obtain texture information, assign to a plurality of areas of the image probability-values reflecting the likelihood that the respective area relates to a specific structure (12), assume at least one geometric property of said specific structure (12), and establish a most possible position parameter of said specific structure taking into account said probability-values and the assumed geometric property; and to provide a steering signal in accordance with the position parameter thus established.