Research on Curvilinear Motion of Automobile with the Application of On-Board Can Bus Data

Sar, Hubert; Brukalski, Mateusz; Rokicki, Krzysztof · 2021 · DOAJ

DOI: 10.26552/com.C.2021.3.B211-B218

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This study addresses the challenge of modeling vehicle curvilinear motion for automotive safety research, specifically focusing on the trade-off between model complexity and parameter identification difficulty. While sophisticated models with high degrees of freedom offer precision, they require numerous parameters that are difficult to obtain. The authors propose using a simpler flat vehicle model validated against real-world data acquired from the vehicle’s on-board Controller Area Network (CAN) bus, offering a cost-effective alternative to professional data acquisition systems. The experimental design involved a road test using a Hyundai Veloster, where signals such as steering wheel angle, lateral acceleration, and yaw rate were recorded via a CAN bus data acquisition card. Vehicle velocity was captured separately using a 10 Hz GPS receiver. The mathematical model employed was a flat vehicle representation requiring only inertial parameters (mass, center of gravity location, yaw mass moment of inertia) and side slip characteristics. The yaw mass moment of inertia was estimated at 2100 kg·m². The simulation utilized random steering wheel angle inputs recorded during the test. Two distinct tire models were applied to describe the relationship between lateral force and side slip angle: a linear model using cornering stiffness coefficients and a non-linear model based on Pacejka’s Magic Formula. The results compared simulated outputs with measured data over a 14-second duration. For lateral acceleration, the linear model yielded a standard deviation of difference of 1074%, whereas the non-linear Magic Formula significantly improved accuracy, reducing the deviation to 281%. Similarly, for yaw angular velocity, the linear model produced a deviation of 845%, while the Magic Formula reduced this to 591%. Visual analysis of the time courses indicated that the linear model overestimated yaw velocity, while the non-linear model slightly underestimated it, but both provided closer alignment with measured values than the linear approach. The study concludes that while CAN bus data acquisition is not a substitute for dedicated professional equipment due to potential phase shifts and signal noise, it provides a viable, low-cost method for validating simple vehicle motion models. The findings demonstrate that incorporating non-linear tire characteristics via Pacejka’s Magic Formula substantially enhances simulation accuracy compared to linear approximations. The authors suggest future work should focus on optimizing parameter identification using methods like neural networks or Monte Carlo algorithms and expanding the model to include roll and pitch degrees of freedom to account for changes in vertical wheel load.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• steering pattern