Vehicle Dynamics Modeling for Accurate State Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current vehicle dynamics models used in self-driving technologies often lack accuracy due to incomplete or inaccurate characteristic parameters of vehicle systems, leading to errors in predicting vehicle state information, which can impact safety.
Innovation Solution
A method and apparatus for constructing a vehicle dynamics model by obtaining sample historical state information and control parameter sequences, adjusting model parameters based on loss values, and training a recurrent neural network model to learn relationships between state information and control parameters, resulting in a more accurate and vehicle-specific model.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If simulation software CarSim is used to construct vehicle dynamics model, then the model construction process is standardized, but the prediction accuracy deteriorates due to incomplete characteristic parameters and manufacturing variations
Solution Approach 1:
The patent transforms the vehicle dynamics model construction from using fixed manufacturer parameters to using dynamically learned parameters from actual vehicle data. The neural network model learns optimal parameter values through training on real vehicle state information, thereby adapting to manufacturing variations and improving prediction accuracy while maintaining standardized construction processes.
Solution Approach 2:
The patent creates a digital copy of the actual vehicle behavior through neural network training. Instead of relying on theoretical manufacturer specifications, the system learns a copy of the true vehicle dynamics characteristics from real operational data, capturing manufacturing variations and actual performance patterns.
2Measurement precision
If heuristic parameter adjustment is performed continuously, then the error between simulation and true vehicle dynamics is reduced, but the time and complexity of model construction increases
Solution Approach 1:
The patent replaces the manual heuristic parameter adjustment process with an automated neural network training system. Instead of requiring continuous manual tuning of parameters, the system automatically learns optimal parameters through machine learning algorithms trained on vehicle data, significantly reducing construction time while maintaining or improving accuracy.
Solution Approach 2:
The vehicle dynamics model performs self-adjustment through automated neural network training. The system automatically learns and optimizes parameters from vehicle data without requiring continuous manual intervention, enabling the model to self-calibrate and adapt to actual vehicle characteristics.
3Device complexity
If manufacturer characteristic parameters are obtained, then the initial model construction is simplified, but errors remain due to manufacturing process variations and part loss
Solution Approach 1:
The patent implements a feedback mechanism where the neural network model is trained on actual vehicle operational data and continuously refines its parameters. The learned parameters from real vehicle performance feed back into the model, correcting errors introduced by manufacturing variations and creating an accurate representation of the specific vehicle instance.
Data Source
AI summary
An embodiment of the present disclosure provides a method and an apparatus for constructing a vehicle dynamics model and a method and an apparatus for predicting vehicle state information. The method of constructing a vehicle dynamics model includes: obtaining sample historical state information and a sample control parameter sequence corresponding to each sample time of a target vehicle and label vehicle state information of each sample time; inputting the sample historical state information and the sample control parameter sequence corresponding to the sample time into an initial vehicle dynamics model to determine sample prediction state information; by using the sample prediction state information and the label vehicle state information, determining a current loss value; based on the current loss value, adjusting model parameters of the initial vehicle dynamics model until the initial vehicle dynamics model reaches a preset convergence state so as to obtain a pre-constructed vehicle dynamics model.