This paper explores the integration of neural differential equations into a physics-guided vehicle single track model. The study demonstrates a significant improvement in model accuracy by reducing the sum of squared error by 68% compared to traditional physics-based models. The research highlights the potential of combining physics-based modeling with machine learning techniques to enhance predictive capabilities in vehicle dynamics. By leveraging advanced sensing and connectivity, data-driven approaches show promise in optimizing vehicle dynamics models for real-time applications. The study compares various modeling methods, including white box, black box, and hybrid models, showcasing the benefits of hybrid modeling in improving accuracy while reducing training data requirements. Through experiments and simulations, the authors emphasize the importance of incorporating physical laws into neural networks to enhance model performance.
To Another Language
from source content
arxiv.org
Key Insights Distilled From
by Stephan Rhod... at arxiv.org 03-19-2024
https://arxiv.org/pdf/2403.11648.pdfDeeper Inquiries