Agricultural Machine Attachment Recognition via 3D Projection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current agricultural machines, such as self-propelled harvesters, face challenges in accurately determining and controlling the components of their attachments, especially under adverse environmental conditions like dust and varying lighting, which affects the precision and reliability of crop flow monitoring and processing.
Innovation Solution
The implementation of a control device with a sensor unit and image processing unit that uses a two-dimensional projection of a three-dimensional model to recognize and identify image areas related to the attachment, allowing for improved component differentiation and process sequence regulation, even when components are not optically detectable or are obscured by crop flow.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If optical sensors are used to detect crop flow and attachment components, then real-time monitoring capability is improved, but detection accuracy deteriorates under adverse environmental conditions like dust and varying lighting
Solution Approach 1:
The patent transitions from two-dimensional image data to three-dimensional point cloud data by introducing LIDAR technology. This dimensional change enables accurate detection of attachment components and crop flow characteristics even under adverse environmental conditions, as the depth information from LIDAR penetration through dust and varying lighting provides robust spatial data that overcomes the limitations of optical cameras
2Reliability
If multiple sensor types are integrated to improve detection reliability, then measurement accuracy is improved, but device complexity increases
Solution Approach 1:
The patent combines LIDAR and optical camera sensors into an integrated sensor system with a unified control unit. This merging approach maintains high detection reliability by utilizing complementary strengths of both sensor types while reducing overall system complexity through integrated data processing and coordinated operation under single control
3Measurement precision
If three-dimensional models are used to recognize image areas, then component identification accuracy is improved, but data processing time increases
Solution Approach 1:
The patent pre-generates three-dimensional models of attachment components before operation and stores them in the control unit. During operation, the system performs template matching between pre-stored models and real-time LIDAR point cloud data, which significantly reduces processing time compared to generating models in real-time, while maintaining high identification accuracy through precise model matching
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The present invention relates to an agricultural machine (1), in particular a self-propelled harvester. The present invention is based on the general concept of recognizing an image area (15), in particular at least one image area or several image areas, with respect to the attachment (2) by means of a two-dimensional projection of a three-dimensional model of an attachment (2) in the images (14).