Generalized Data Generation for Road Surface Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Estimating road surface conditions using machine learning requires extensive training data for each specific type of surface, such as smooth and rough roads, leading to increased data requirements and inefficiencies.
Innovation Solution
A generalized data generation and estimation system that trains a model using a general dataset with a general parameter, allowing it to generate and process input datasets with various parameters, reducing the need for extensive training data by converting data into a standardized format for estimation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If training data is collected for each specific road surface type (smooth, rough, etc.), then estimation accuracy for each type is improved, but the amount of training data required increases significantly
Solution Approach 1:
The patent applies universality by training a single machine learning model on generalized road surface data that encompasses multiple surface types (smooth, rough, etc.). Instead of creating separate specialized models for each road surface type, one universal model is trained on diverse data representing various conditions, enabling it to accurately estimate all types of road surfaces without requiring separate training datasets for each type.
2Measurement precision
If separate trained models are created for each road surface type, then estimation accuracy is improved, but device complexity increases
Solution Approach 1:
The patent applies merging by combining multiple specialized models into a single unified model. Instead of maintaining separate trained models for smooth roads, rough roads, and other surface types, the approach merges these into one machine learning model that is trained on comprehensive generalized data covering all road surface conditions, thereby reducing system complexity while maintaining estimation accuracy.
3Adaptability or versatility
If extensive training data for multiple road surface types is collected, then comprehensive coverage is improved, but data generation time and cost increase
Solution Approach 1:
The patent applies parameter changes by transforming the approach from collecting extensive real-world data for each road surface type to generating generalized training data through parameter variations. By using synthesized data with varied parameters representing different road conditions (smooth, rough, wet, dry, etc.), the system achieves comprehensive road surface coverage without the time-consuming process of collecting and labeling extensive real-world datasets for each condition.
Data Source
AI summary
A generalized data generation device and estimation device, a generalized data generation method and estimation method, and a generalized data generation program and estimation program are provided, which are capable of estimating the state of a target object accurately with the amount of training data being reduced. The generalized data generation device 10 includes: a training unit 101 that trains a generalized model for training 141 for obtaining data satisfying a general parameter through predetermined machine learning by using a general training dataset 142 as input, which is a set of data satisfying the general parameter from among multiple types of parameters, and outputs a trained generalized model 143; and a generalized data generation unit 102 that generates a generalized input dataset 145 generalized by using an input dataset 144, which is a set of data satisfying any of the multiple types of parameters, and the trained generalized model 143, such that the input dataset 144 satisfies the general parameter.