Work Implement Position Estimation Using Trained Model
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing configuration for determining the posture of a work implement in hydraulic excavators requires multiple angle sensors, increasing the number of components and complexity.
Innovation Solution
A system that uses a computer-implemented method to estimate the position of a work implement's target portion using a trained position estimation model, based on operation command values and displacement data, eliminating the need for explicit angle sensors by leveraging operation command values and displacement data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If angle sensors are attached to the boom, dipper stick, and bucket to determine work implement posture, then positional accuracy is improved, but device complexity and component count increase
Solution Approach 1:
The patent replaces mechanical angle sensors with a computational approach using operation command values and a trained position estimation model (neural network) to estimate work implement position, thereby eliminating physical sensors while maintaining measurement capability
Solution Approach 2:
The patent creates a virtual model (position estimation model) that replicates the functionality of physical angle sensors by learning the relationship between operation commands and actual positions through training data, allowing position estimation without direct physical measurement
2Measurement precision
If multiple angle sensors are installed to measure work implement angles, then measurement precision is improved, but device complexity increases
Solution Approach 1:
The patent substitutes mechanical angle sensing systems with an information processing system that uses operation command values as input to a trained neural network model, eliminating the need for physical angle sensors while achieving comparable or superior measurement precision
Solution Approach 2:
The patent introduces operation command values as an intermediary parameter that indirectly represents work implement position, using these command values as input to the position estimation model rather than directly measuring physical angles
Data Source
AI summary
There is provided a system including a work machine, comprising a main body of the work machine, a work implement attached to the main body of the work machine, an operating member operated to operate the work implement, and a computer. The computer has a trained position estimation model for determining the position of a portion of the work implement that is a target of positional estimation. The computer obtains an operation command value causing the work implement to make a movement in response to an operation of the operating member, uses the trained position estimation model to estimate from the operation command value an amount of displacement of the target of positional estimation from a reference position, and outputs an estimated position that is a position of the target of positional estimation estimated from the reference position and the amount of displacement.