Residual Learning towards High-fidelity Vehicle Dynamics Modeling with Transformer
Authors:
- Jinyu MiaoJMJinyu Miao
- Rujun YanRYRujun Yan
- Bowei ZhangBZBowei Zhang
- Tuopu WenTWTuopu Wen
- Kun JiangKJKun Jiang
- Mengmeng YangMYMengmeng Yang
- Jin HuangJHJin Huang
- Zhihua ZhongZZZhihua Zhong
- Diange YangDYDiange Yang
Preprint
Abstract
The vehicle dynamics model serves as a vital component of autonomous driving systems, as it describes the temporal changes in vehicle state. In a long period, researchers have made significant endeavors to accurately model vehicle dynamics. Traditional physics-based methods employ mathematical formulae to model vehicle dynamics, but they are unable to adequately describe complex vehicle systems due to the simplifications they entail. Recent advancements in deep learning-based methods have addressed this limitation by directly regressing vehicle dynamics. However, the performance and generalization capabilities still require further enhancement. In this letter, we address these problems by proposing a vehicle dynamics correction system that leverages deep neural networks to correct the state residuals of a physical model instead of directly estimating the states. This system greatly reduces the difficulty of network learning and thus improves the estimation accuracy of vehicle dynamics. Furthermore, we have developed a novel Transformer-based dynamics residual correction network, DyTR. This network implicitly represents state residuals as high-dimensional queries, and iteratively updates the estimated residuals by interacting with dynamics state features. The experiments in simulations demonstrate the proposed system works much better than physics model, and our proposed DyTR model achieves the best performances on dynamics state residual correction task, reducing the state prediction errors of a simple 3 DoF vehicle model by an average of 92.3% and 59.9% in two dataset, respectively.Keywords
All Related Versions
This publication has 15 references indexed in Scilit:
- Deep Dynamics: Vehicle Dynamics Modeling With a Physics-Constrained Neural Network for Autonomous RacingIEEE Robotics and Automation Letters, 2024
- Poses as Queries: End-to-End Image-to-LiDAR Map Localization With TransformersIEEE Robotics and Automation Letters, 2023
- Online learning of MPC for autonomous racingRobotics and Autonomous Systems, 2023
- Nonlinear System Identification: A User-Oriented Road MapIEEE Control Systems, 2019
- Learning-Based Model Predictive Control for Autonomous RacingIEEE Robotics and Automation Letters, 2019
- Dynamic Modeling and Control of High-Speed Automated Vehicles for Lane Change ManeuverIEEE Transactions on Intelligent Vehicles, 2018
- Mastering the game of Go with deep neural networks and tree searchNature, 2016
- Cornering stiffness and sideslip angle estimation based on simplified lateral dynamic models for four-in-wheel-motor-driven electric vehicles with lateral tire force informationInternational Journal of Automotive Technology, 2015
- A Switching Rollover Controller Coupled With Closed-Loop Adaptive Vehicle Parameter IdentificationIEEE Transactions on Intelligent Transportation Systems, 2014
- Optimal control for a Formula One car with variable parametersVehicle System Dynamics, 2013