Detail View

Incorporating Road Topography and Bank Angle Estimation in Model Predictive Control for Enhancing Path Tracking Accuracy and Stability of Autonomous Vehicles
Citations

WEB OF SCIENCE

Citations

SCOPUS

Metadata Downloads

Title
Incorporating Road Topography and Bank Angle Estimation in Model Predictive Control for Enhancing Path Tracking Accuracy and Stability of Autonomous Vehicles
Alternative Title
자율주행 차량의 경로 추종 정확도와 안정성 향상을 위한 모델 예측 제어에서 도로 지형 통합 및 경사각 추정
DGIST Authors
Jeongmin ChoiYongseob LimKanghyun Nam
Advisor
임용섭
Co-Advisor(s)
Kanghyun Nam
Issued Date
2025
Awarded Date
2025-02-01
Citation
Jeongmin Choi. (2025). Incorporating Road Topography and Bank Angle Estimation in Model Predictive Control for Enhancing Path Tracking Accuracy and Stability of Autonomous Vehicles. doi: 10.22677/THESIS.200000843789
Type
Thesis
Description
Road Topography, Model Predictive Control, Road Bank, Extended Kalman Filter with Unknown Inputs, Long Short-Term Memory
Abstract
In recent years, incorporating road topography into autonomous vehicle control strategies has gained importance due to its impact on vehicle stability and performance. This study verifies the performance of Model Predictive Control (MPC) by integrating road curvature and bank angle into the control strategy. A Nonlinear Model Predictive Control (NLMPC) framework is employed for path tracking, using two different methods for estimating road bank angle: the Extended Kalman Filter with Unknown Input (EKF-UI) and a Long Short-Term Memory (LSTM) neural network-based estimator. The LSTM estimator, expected to handle more complex nonlinearities, is compared with the EKF-UI in terms of estimation accuracy and stability in dynamic environments. To further enhance lateral stability, side slip constraints are incorporated using slack variables, and the Zero-Moment-Point (ZMP) is included as a standard constraint to improve stability by accounting for the bank angle. Simulations in MATLAB and CarSim validate the proposed control strategy, showing that the LSTM-based estimator significantly improves stability and accuracy in path tracking, especially in complex road conditions, outperforming the EKF-UI. In conclusion, the model incorporating road curvature and bank angle, particularly with the LSTM estimator, is proven to be more effective and reliable in dynamic scenarios. |최근 몇 년간 도로 지형을 자율주행 차량 제어 전략에 통합하는 것이 차량의 안정성과 성능에 미치는 영향 때문에 중요성이 커지고 있습니다. 본 연구는 도로 곡률과 뱅크 각을 제어 전략에 통합하여 모델 예측 제어(MPC)의 성능을 검증했습니다. 비선형 모델 예측 제어(NLMPC) 프레임워크가 경로 추종에 사용되었으며, 도로 뱅크 각 추정을 위해 두 가지 방법, 즉 미지 입력 확장 칼만 필터(EKF-UI)와 장단기 기억(LSTM) 신경망 기반 추정기를 사용했습니다. 더 복잡한 비선형성을 처리할 것으로 예상되는 LSTM 추정기는 EKF-UI와 비교하여 동적 환경에서의 추정 정확도와 안정성을 평가했습니다. 횡방향 안정성을 더욱 향상시키기 위해 슬랙 변수를 사용하여 사이드슬립 제약을 포함하였으며, 뱅크 각을 고려하여 안정성을 개선하기 위해 제로 모멘트 포인트(ZMP)를 표준 제약으로 포함시켰습니다. MATLAB과 CarSim에서의 시뮬레이션을 통해 제안된 제어 전략의 성능을 검증한 결과, LSTM 기반 추정기가 특히 복잡한 도로 조건에서 경로 추종의 안정성과 정확성을 크게 향상시켜 EKF-UI를 능가하는 성능을 보였습니다. 결론적으로, 도로 곡률과 뱅크 각을 통합하고 특히 LSTM 추정기를 활용한 모델이 동적 환경에서 더 효과적이고 신뢰할 수 있음을 입증했습니다.
Table Of Contents
List of Contents
Abstract i
List of contents ii
List of tables iii
List of figures iv

I. Introduction 4
II. Related Works 7
III. System Design
3.1 Nonlinear Vehicle Dynamic Model 10
3.2 Extended Kalman Filter Based Vehicle State Estimator 14
IV. Controller Design
4.1 Driving Safety Constraints for Vehicle Stability Control 16
4.2 NLMPC controller design 19
4.3 PI Control for Longitudinal Velocity Command Generation 22
V. Road Bank Estimation
5.1 EKF-UI Algorithm-based Road Bank Estimation 24
5.2 LSTM-based Road Bank Estimation 26
VI. Simulation
6.1 Simulation Scheme 28
6.2 Simulation Results
6.2.1 Results at 60km/h 31
6.2.2 Results at 80km/h 37
VII. Conclusion 44
URI
http://hdl.handle.net/20.500.11750/58025
http://dgist.dcollection.net/common/orgView/200000843789
DOI
10.22677/THESIS.200000843789
Degree
Master
Department
Department of Robotics and Mechatronics Engineering
Publisher
DGIST
Show Full Item Record

File Downloads

  • There are no files associated with this item.

공유

qrcode
공유하기

Total Views & Downloads