295-299 2026 [춘계학술대회]
| 제목 | 차량 한계 주행 조건을 고려한 실시간 딥러닝-EKF 기반 하이브리드 사이드 슬립각 추정기 개발 |
|---|---|
| 분야 | 지능형 샤시 및 제어 |
| 언어 | Korean |
| 저자 | 정해영(경상국립대학교), 박채은(경상국립대학교), 오병규(경상국립대학교), 이승우(경상국립대학교), 한상원(경상국립대학교) |
| Key Words | Side Slip Angle(사이드 슬립 각), Deep Learning(딥러닝), Extended Kalman Filter(확장 칼만 필터), Vehicle Dynamics(차량 동역학), Vehicle Stability(차량 주행 안정성), State Estimation(상태 추정) |
| 초록 |
Recent advancements in high-performance vehicles and autonomous driving technologies have increased the importance of accurately analyzing vehicle lateral dynamics. Among key variables, the side slip angle is a critical factor in evaluating vehicle stability, especially during cornering and sudden steering inputs. However, direct measurement of the side slip angle is not feasible in production vehicles, and estimation methods based on IMU sensors are commonly used. In this paper, a hybrid estimation framework combining an Extended Kalman Filter (EKF) and a deep learning-based lateral velocity estimator is proposed to improve estimation accuracy under nonlinear conditions. An LSTM (Long Short-Term Memory) network is designed using chassis-related signals, and its predicted states are incorporated into the EKF as additional measurements. The proposed method is validated through CarMaker–MATLAB/Simulink co-simulation under various driving conditions. The results demonstrate stable and reliable estimation performance, even in nonlinear and low-friction scenarios, with improved accuracy indicated by reduced RMSE and MAE. |
| 원문(PDF) | 다운로드 |