사륜 조향 차량의 횡안정성 향상을 위한 LPV-MPC 및 임베디드 PiLS 실시간 검증 연구

초록

This paper presents a linear parameter-varying model predictive control (LPV-MPC) strategy for enhancing the lateral stability of four-wheel steering (4WS) vehicles. Its real-time performance validated for the first time in an embedded Processor-in-the-Loop Simulation (PiLS) environment using MATLAB/Simulink and a Teensy 4.1 development board. The proposed controller is benchmarked against front-wheel steering (FWS), linear quadratic regulator (LQR), and uncontrolled cases under two representative scenarios: step steering and double-lane change maneuvers. In the step steering test, performance is evaluated in terms of maximum deviation and settling time, while in the double-lane change test, root mean square error (RMSE) is used as the primary performance metric. PiLS results confirm that all computations are completed within the sampling period, demonstrating the feasibility of real-time embedded implementation. Compared to the uncontrolled baseline, the LPV-MPC-based 4WS controller achieves a 53.6% reduction in yaw rate RMSE and outperforms the LQR controller by 65.3% in double-lane change maneuvers while also reducing the peak-to-peak sideslip angle by 25.7%. In the step steering test, it achieves a 1.2% lower maximum deviation than FWS with a fast settling time of 3.65 s.

키워드