CAJ | 학술논문

小车自主导航系统是路径规划领域研究的难点问题。本文利用自适应动态规划算法对小车自主寻路、避障的控制策略进行了设计。介绍了仿真过程中小车检测环境状态量的方法，提出了小车通过连续型奖励/惩罚信号自主学习寻路、避障的控制策略，并结合虚拟现实技术创建了不同路径和不同障碍物状态的仿真环境。通过不同路径状态的仿真实验，验证了自适应动态规划算法在解决小车自主导航避障控制策略问题上的可靠性和稳定性。
소차자주도항계통시로경규화영역연구적난점문제。본문이용자괄응동태규화산법대소차자주심로、피장적공제책략진행료설계。개소료방진과정중소차검측배경상태량적방법，제출료소차통과련속형장려/징벌신호자주학습심로、피장적공제책략，병결합허의현실기술창건료불동로경화불동장애물상태적방진배경。통과불동로경상태적방진실험，험증료자괄응동태규화산법재해결소차자주도항피장공제책략문제상적가고성화은정성。
The car autonomous navigation system has been a hot topic in path planning. This paper focuses on control strategy design for car autonomous navigation with obstacles avoidance by using adaptive dynamic programming (ADP) approach. First, the method for car observing the environment in simulation process is introduced. Second, a continuous reward/punishmen signal for car autonomous learning the control strategy is proposed. Third, the different simulation environment with different obstacles' states are established by virtual reality (VR) technology. The simulation results show that the ADP approach provides the suitable reliability and stability for control strategy design on autonomous navigation problem.