基于移动子目标的复合式路径规划算法  

Compound Path Planning Algorithm Based on Sliding Subtarget

作　　者：张嘉琦[1] 

机构地区：[1]长安大学公路学院,陕西西安710064

出　　处：《中国公路学报》2017年第11期138-146,共9页China Journal of Highway and Transport

基　　金：国家自然科学基金项目(61503043);陕西省自然科学基础研究计划项目(2015JQ6214)

摘　　要：为使无人驾驶汽车无碰撞、高效到达目标点,针对当前未同时突出全局路径和局部路径优点的问题,通过激光传感器检测无人驾驶汽车周围环境,用子目标点与无人驾驶汽车的连线将平面环境划分为左、右2个半平面,分别搜索可通行的自由扇区,根据一种代价函数评价各个自由扇区得到最优的路径。考虑无人驾驶汽车各时刻的姿态不同,为无碰撞的切入最优路径,创建基于无人驾驶汽车到阻塞区域距离和当前车速的风险评价函数,基于人的驾驶行为,创建以风险评价值和切入最优扇区的转角为输入,无人驾驶汽车角速度和线速度为输出的模糊控制。在全局路径已知的条件下,在全局路径上为实时的局部路径规划选择子目标点。根据环境等因素的影响,将子目标划分成被未知阻塞区域遮挡、被已知阻塞区域遮挡且遮挡前子目标可见、被已知阻塞区域遮挡且遮挡前子目标不可见3种状态,为引导无人驾驶汽车,分别分析其子目标所处的状况并提出子目标应当按照一定速度向特定方向移动的方案。研究结果表明:该方法能够很好地解决局部路径规划层无远见的问题,全局路径上的子目标点有效地引导无人驾驶汽车前往最终的目标,同时又不会与阻塞区域发生碰撞,保证了无人驾驶汽车行驶的平顺性,实现实时无人驾驶汽车路径达到最优,又很好地关联了全局路径规划层和局部路径规划层。In order to take autonomous vehicles to the destination efficiently with no collision,the problem of impossibility of highlighting advantages of both global and local paths at the same time was deduced.The environment around autonomous vehicles was detected by dint of the laser sensor,whilst the plane environment was divided into left and right half planes by the connection of sub-target point and autonomous vehicles.After searching and evaluating the accessible free sectors based on a cost function,the optimal paths were selected.In view of the different vehicle conditions at different time,a risk assessment function was established based on the distance between congested area and the autonomous vehicle,and the current vehicle speed to guarantee the optimal path with no collision.The fuzzy control was established,on the basis of driving behaviors,regarding the risk assessment value and angle into the optimal sector as the input,and angular velocity and the linear velocity of vehicles as the output.Meanwhile,with the global path known,the sub-target was selected for the real-time local path planning on the global path.Onaccount of the influence of environment and other factors,the sub-target was classified into the place blocked by the unknown area,somewhere obstructed by the known congested area with visible sub-target and somewhere blocked by known area with invisible sub-target.In order to guide the autonomous vehicles,the state of each sub-target was analyzed respectively,and vehicles should move forward at the certain speed in a certain direction.The results show that the proposed method tends to solve the problem of no foresight in the local path planning.The subtarget on the global path can effectively guide the vehicle to the final target without any collision,sheltering the vehicle from obstacles.Such a method ensures the optimal real-time path of autonomous vehicles,and correlates the global path planning layer with the local path planning layer.

关 键 词：交通工程 无人驾驶汽车 路径规划 子目标 模糊控制 

分 类 号：U491.12[交通运输工程—交通运输规划与管理]