Vehicle Trajectory Specification in Presence of Traffic Lights with Known or Uncertain Switching Times
Vehicle Trajectory Specification in Presence of Traffic Lights with Known or Uncertain Switching Times
复制标题
存在已知或不确定切换时间的交通灯时的车辆轨迹规范
DOI：
--
复制
发表时间：
2020
影响因子：
1.7
通讯作者：
M. Papageorgiou
中科院分区：
文献类型：
--
作者：
P. Typaldos;Ioanna Kalogianni;K. Mountakis;I. Papamichail;M. Papageorgiou
这项工作的主要目的是生成最佳轨迹的车辆穿越信号交叉口，与交通信号操作在固定时间或实时（自适应）模式。在后一种情况下，下一个切换时间是基于当前的交通状况在真实的时间中决定的，因此是预先不确定的。GLOSA（绿色光最优速度咨询）问题通过使用交通信号灯信息并且基于车辆的初始状态（位置和速度）和固定的最终目的地状态计算车辆的轨迹和速度轮廓来解决。首先，一个适当的最优控制问题，制定和解决解析通过庞特里亚金的最小值原理（PMP）的情况下，已知的开关时间。随后，对于实时信号的情况下，可能的信号切换时间的时间窗口的可用性，沿着与相应的概率分布，被假定，并且该问题以随机最优控制问题的格式被铸造，并且使用随机动态规划（SDP）技术数值地解决。应用结果，为各种驾驶情况下，确定性的方法，考虑已知的切换时间的情况下，和一个全面的比较随机GLOSA方法与次优方法。特别是，它表明，所提出的SDP方法实现了更好的平均性能相比，次优的方法，因为更好的（概率）信息的交通灯切换时间。
The main purpose of this work is to generate optimal trajectories for vehicles crossing a signalized junction, with traffic signals operated in either fixed-time or real-time (adaptive) mode. In the latter case, the next switching time is decided in real time based on the prevailing traffic conditions and is therefore uncertain in advance. The GLOSA (Green Light Optimal Speed Advisory) problem is addressed by using traffic lights information and calculating a trajectory and velocity profile for the vehicle based on the vehicle’s initial state (position and speed) and a fixed final destination state. At first, an appropriate optimal control problem is formulated and solved analytically via Pontryagin’s minimum principle (PMP) for the case of known switching times. Subsequently, for the case of real-time signals, availability of a time-window of possible signal switching times, along with the corresponding probability distribution, is assumed, and the problem is cast in the format of a stochastic optimal control problem and is solved numerically using stochastic dynamic programming (SDP) techniques. Application results, for various driving scenarios, of the deterministic approach, which considers the case of known switching times, and a comprehensive comparison of the stochastic GLOSA approach with a sub-optimal approach are presented. In particular, it is demonstrated that the proposed SDP approach achieves better average performance compared with the sub-optimal approach because of the better (probabilistic) information on the traffic light switching time.