摘要
在自动驾驶技术普及的过程中,车辆与道路行人之间的避撞方法一直是人们关注的热点话题。本文将行人运动的不确定性、运动车辆对行人运动的影响等因素与熵的概念相结合,提出一种基于可行驶区域的自动驾驶避撞方法,以自定义的车辆可行驶区域与对应空间熵的计算为基础,再应用于车辆与行人之间的碰撞预测。首先,利用车辆自身传感器获取的信息数据,采用速度与路径规划叠加的方法来生成车辆备选的轨迹路线;其次,基于社会力模型与马尔可夫模型对行人位置进行预测,得到行人位置概率;最后,根据目标优化算法得到车辆行驶的最优轨迹,实现避撞。仿真实验表明,本文提出的基于可行驶区域的自动驾驶避撞方法在不同的行车工况下可以有效地与行人进行安全避撞交互,对于推动自动驾驶的发展、保障道路行人安全具有重要意义。
In the process of the popularization of autonomous driving technology,the collision avoidance method between vehicles and pedestrians has always been a hot topic.In this paper,the uncertainty of pedestrian movement,the influence of moving vehicles on pedestrian movement and other factors are combined with the concept of entropy,and an automatic driving collision avoidance method based on driving zone is proposed.This method is based on the calculation of self-defined vehicle driving area and corresponding space entropy,and then applied to the collision prediction between vehicles and pedestrians.First,the information data obtained from the vehicle′s own sensor is used to generate the alternative trajectory route of the vehicle by the method of superposition of speed and path planning.Then the pedestrian position is predicted based on the social force model and Markov model to obtain the pedestrian position probability.Finally,the optimal trajectory of the vehicle is obtained according to the objective optimization algorithm to achieve collision avoidance.Simulation experiments show that the proposed automatic driving collision avoidance method based on driving zone can effectively interact with pedestrians safely under different driving conditions,which is of great significance to promote the development of automatic driving and ensure the safety of pedestrians on the road.
作者
武鹏柯
张伟伟
武建龙
程家镯
WU Pengke;ZHANG Weiwei;WU Jianlong;CHENG Jiazhuo(School of Mechanical and Automotive Engineering,Shanghai University of Engineering Science,Shanghai 201620,China)
出处
《智能计算机与应用》
2023年第12期107-113,共7页
Intelligent Computer and Applications
关键词
自动驾驶
可行驶区域
空间熵
避撞
目标优化
autonomous driving
driving area
spatial entropy
collision avoidance
objective to optimize