조향계의 동적특성을 포함한 예측 경로 추종 제어

Abstract

Path tracking control is one of the most important functions for autonomous driving. In path tracking control, high accuracy and smooth tracking are required for safe and comfort driving. In order to meet these requirements, model predictive control approaches, which can obtain an optimized solution with respect to a predefined path, have been widely used. Conventional predictive controllers have been researched based on a simple bicycle model. However, the conventional predictive controllers have a performance limitation in practical challenges due to the difference between the simple bicycle model and the actual vehicle. To overcome this limitation, the actuator dynamics of the steering system should be incorporated into the control design. In this paper, a model predictive control based path tracking control algorithm is proposed to achieve the accurate and smooth tracking by incorporating the dynamic characteristics of the steering actuation system. In the proposed control algorithm, an optimal trajectory of the steering command is calculated by applying a quadratic programming optimization method. The proposed controller was verified by computer simulation with various driving scenarios. The simulation results show that the proposed controller can improve the tracking performance.|경로추종제어는 자율주행에 필수적인 요소이다. 주행 안전과 승차감을 위해서 경로 추종제어에서는 정확하고 부드러운 추종을 요구하며, 이 요구조건을 충족시키기 위해서 주어진 경로에 대하여 최적의 해법을 얻는 모델예측제어기법이 널리 사용되어 왔다. 기존의 모델예측제어기반 경로추종제어기들은 제어 설계의 편의성을 위해 간단한 자전거 모델을 기반으로 연구되었다. 하지만 실제 자율주행에 있어서 간단한 자전거 모델은 실제 차량과 큰 차이가 존재하기 때문에 기존의 제어시스템들은 추종성능 향상에 한계를 갖게 된다. 이러한 차이를 줄이기 위해서는 조향시스템의 동적 특성이 제어시스템에 고려되어야 하며, 제어시스템은 이 동적 특성을 보상할 수 있어야 한다. 이 논문에서는 정확하고 부드러운 추종을 위해 자동조향시스템의 동적 특성을 포함한 모델예측제어 기반의 경로추종제어 기법이 제안되었다. 제안된 제어기는 최적조향입력 값을 이차 계획법(quadratic programming)을 적용함으로써 얻을 수 있었다. 제안된 제어기는 다양한 주행 시나리오와 함께 컴퓨터 시뮬레이션을 통하여 검증되었으며, 시뮬레이션 결과를 통해 제안된 제어기가 추종성능을 향상시킬 수 있다는 것을 확인 하였다.; Path tracking control is one of the most important functions for autonomous driving. In path tracking control, high accuracy and smooth tracking are required for safe and comfort driving. In order to meet these requirements, model predictive control approaches, which can obtain an optimized solution with respect to a predefined path, have been widely used. Conventional predictive controllers have been researched based on a simple bicycle model. However, the conventional predictive controllers have a performance limitation in practical challenges due to the difference between the simple bicycle model and the actual vehicle. To overcome this limitation, the actuator dynamics of the steering system should be incorporated into the control design. In this paper, a model predictive control based path tracking control algorithm is proposed to achieve the accurate and smooth tracking by incorporating the dynamic characteristics of the steering actuation system. In the proposed control algorithm, an optimal trajectory of the steering command is calculated by applying a quadratic programming optimization method. The proposed controller was verified by computer simulation with various driving scenarios. The simulation results show that the proposed controller can improve the tracking performance.