Field Route Localization by Image Segmentation for Agricultural Machines
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing GPS-based systems for locating agricultural machines are costly due to satellite data license fees and inefficient in data processing, posing security and efficiency challenges.
Innovation Solution
A device that uses sensor data, such as camera images, to localize agricultural machines by comparing monitored areas with predefined routes, eliminating the need for GPS and reducing data transmission and processing costs, while enhancing accuracy through correlation values and semantic segmentation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If GPS-based systems are used for locating agricultural machines, then positioning accuracy is improved, but system cost and data transmission requirements worsen
Solution Approach 1:
The patent creates a digital copy of the field environment using sensor data (images, point clouds) and compares it with a predefined digital route model. This copying approach replaces GPS satellite signal dependency with local environmental data matching, achieving high positioning accuracy without expensive satellite licensing fees.
Solution Approach 2:
The patent introduces an intermediary evaluation unit that processes sensor data and compares it with the predefined route to determine machine position. This intermediary layer replaces direct GPS satellite communication, eliminating the need for expensive satellite data licenses while maintaining positioning functionality through local environmental feature matching.
2Reliability
If GPS data is transmitted and processed over long distances, then positioning functionality is achieved, but data security and processing efficiency worsen
Solution Approach 1:
The patent extracts positioning functionality from the GPS satellite system and implements it locally using onboard sensors and processing units. By taking out the core positioning function from the external GPS infrastructure, the system eliminates the need to transmit sensitive positioning data over long distances, thereby improving data security while maintaining positioning reliability.
Solution Approach 2:
The agricultural machine performs positioning calculations autonomously using its own sensor data and onboard processing capabilities. The machine serves itself by comparing its sensor-captured environment with the predefined route, eliminating dependency on external GPS data transmission and improving both data security and processing efficiency.
3Productivity
If sensor data is processed locally without GPS, then cost and data transmission are reduced, but positioning accuracy may worsen
Solution Approach 1:
The patent segments the field route into multiple sections and compares sensor data against each segment independently. This segmentation allows for more precise local matching between captured environmental features and the predefined route, maintaining high positioning accuracy while enabling efficient local processing without GPS dependency.
Solution Approach 2:
The patent transitions from two-dimensional GPS coordinate matching to three-dimensional environmental feature matching using sensor data (images, point clouds). By adding dimensional depth through environmental feature comparison, the system achieves superior positioning accuracy locally, replacing GPS precision with more richly-detailed sensor-based spatial matching.
Data Source
Figure 1~2
Figure 3
AI summary
Apparatus for locating an agricultural machine on the basis of sensor data and image segmentation. Apparatus (10) for locating an agricultural machine (50) on an agricultural useful area (60), comprising an input interface (12) for inputting a route (11a) predefined on the useful area (60) and for inputting image data (11b) relating to a monitoring region of the useful area (60), which monitoring region is captured by a sensor (22) and is in the vicinity of the agricultural machine (50), an evaluation unit (14) for detecting a position of the monitoring region with respect to the predefined route (11a), and an output interface (16) for outputting the detected position.