Gaussian Mixture Model Driver Action Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing driver models, such as those using fuzzy rules or neural networks, are difficult to create and do not accurately represent individual driving characteristics, making it challenging to estimate driving actions and evaluate driving conditions effectively.
Innovation Solution
A driving action estimating device that uses a Gaussian mixture model (GMM) to describe the probability distribution of characteristic amounts detected during vehicle travel, allowing for the calculation of maximum posterior probabilities to estimate driving actions and create precise driver models for each driver, including operation amounts and time change data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If driver models are created using fuzzy rules or neural networks, then the model can represent driving characteristics, but the creation process becomes complex and difficult
Solution Approach 1:
The patent replaces complex machine learning systems (fuzzy rules, neural networks) with a statistical mathematics-based approach using probability density functions. This substitution simplifies the model creation process while maintaining the ability to represent individual driving characteristics accurately through parameter estimation from observed driving data.
Solution Approach 2:
The patent changes the fundamental parameters of the driver model from complex structural configurations (neural network weights, fuzzy rule bases) to simple statistical parameters (mean, variance of normal distributions). This parameter transformation enables easy model creation through straightforward statistical estimation rather than complex training procedures.
2Adaptability or versatility
If a general driver model is created, then it can be applied broadly, but it cannot precisely represent individual driving characteristics
Solution Approach 1:
The patent applies local quality by allowing each driver to have their own specific parameter values (mean and variance of normal distributions) while maintaining the same overall model structure. This enables the model to be universally applicable in form but locally adapted in parameters, precisely capturing individual driving characteristics without sacrificing broad applicability.
3Reliability
If driver models are created from collected driving data, then individual characteristics can be captured, but the data may not represent normal driving conditions
Solution Approach 1:
The patent inverts the conventional approach by not trying to filter out abnormal data points, but rather using all observed data to estimate statistical parameters. The probability density function approach naturally handles variability in driving conditions by capturing the overall distribution characteristics, ensuring that even abnormal conditions contribute to understanding the driver's true characteristics.
Data Source
AI summary
A driver model with higher precision is created as an evaluation standard for a driving condition in a normal condition. By detecting biometric information of a driver, whether a driver is in a usual condition or not is recognized. Then, data of driving conditions are collected while the driver is driving, and from the driving condition data, a part indicating that the driver operates in a usual condition is extracted to create a driver model. Further, the driver model is created taking only a case of driving in a normal condition as a driving action in normal times based on biometric information of the driver, and hence the driver model becomes more precise and neutral. Further, by using a GMM (Gaussian mixture model) for the driver model, a driver model for each driver can be created easily, and moreover, by calculation to maximize a conditional probability, a driving operation action is easily estimated and outputted.