Mobile Machine Localization Using Trunk-Based Orchard Mapping
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In environments with high-density tree distributions, such as vineyards or forests, the distribution of tree leaves changes significantly with seasons, making it difficult to maintain accurate positioning using GNSS and challenging to apply SLAM for autonomous mobile machine navigation.
Innovation Solution
A mobile machine equipped with a LiDAR sensor generates an environment map focusing on tree trunks, which undergo less seasonal change, allowing continuous use of the map over time, enabling autonomous movement and automatic steering.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If SLAM is used for autonomous navigation in high-density tree environments, then positioning capability is improved, but the system cannot maintain accurate positioning across seasons due to leaf distribution changes
Solution Approach 1:
The patent extracts only the trunk components from the complete tree structure to create the environment map, excluding the leaves and branches that cause seasonal variations. This selective extraction allows the system to maintain consistent environmental features across different seasons while preserving positioning capability.
Solution Approach 2:
The patent applies different treatment to different parts of the tree: trunks are mapped with high detail and permanence as they remain consistent year-round, while leaves and branches are excluded or treated as transient elements. This local differentiation resolves the contradiction between maintaining positioning accuracy and adapting to seasonal changes.
2Measurement precision
If environment maps are regenerated frequently to account for seasonal changes, then map accuracy is improved, but operational continuity is reduced due to inability to use historical maps
Solution Approach 1:
By extracting only the stable trunk features from the environment, the system creates maps that remain accurate over extended periods without requiring frequent regeneration, thus extending the usable duration of environment maps while maintaining sufficient accuracy for navigation.
3Loss of information
If complete tree structures including leaves are mapped, then initial map detail is improved, but long-term usability deteriorates due to seasonal leaf distribution changes
Solution Approach 1:
The system extracts and maps only the trunk portions of trees, deliberately excluding leaves and branches. This extraction approach preserves sufficient environmental detail for navigation while eliminating the seasonal variability that would otherwise limit map validity periods.
Solution Approach 2:
Instead of mapping complete trees and then filtering out seasonal elements, the patent inverts the approach by directly mapping only the permanent trunk structures from the beginning, thereby achieving both detail retention and long-term usability.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Enables accurate autonomous navigation and steering in environments where GNSS is difficult by utilizing trunk-based mapping, ensuring consistent operation despite seasonal leaf changes.
Implementation Method 1
a sensor unit that repeatedly acquires sensor data indicating a distribution of objects in a surrounding environment of the mobile machine
Data Source
Figure 1
Figure 2A~2B
Figure 3
AI summary
High-accuracy positioning is realized in an environment where multiple trees grow. The mobile machine moves between multiple rows of trees. The mobile machine includes: one or more sensors to output sensor data indicating a distribution of objects in a surrounding environment of the mobile machine; and a data generator. While performing localization, the data generator detects trunks of the rows of trees in the surrounding environment of the mobile machine based on the sensor data that is repeatedly output from the one or more sensors, and generates local map data from which to generate environment map data indicating a distribution of the detected trunks of the rows of trees and record the local map data to a storage device.