The invention discloses a refined path scheduling method for an automatic driving vehicle at an intersection without signal lamps based on a double-learning network. According to the method, an automatic driving vehicle is taken as a research object, grid discretization processing is carried out on a non-signal intersection, and a path learning data set is obtained according to a path planning algorithm. And based on the data set, a coarse-granularity fast path scheduling decision model is constructed through a CNN network, and fast decision scheduling is carried out on the real-time path of the autonomous vehicle at the non-signalized intersection. In order to realize refined path scheduling of an automatic driving vehicle, on the basis of a coarse-granularity fast-degree path scheduling decision model, a DQN reinforcement learning network is added to calculate an evaluation value of a model decision behavior, a reward function is constructed, and an adaptive learning rate is used to improve the accuracy of model prediction. And the real-time performance and the accuracy of path planning and scheduling of the automatic driving vehicle at the non-signalized intersection are balanced. The traffic efficiency of the autonomous vehicle when passing through the intersection without the signal lamp can be improved.
本发明公开了一种基于双学习网络的无信号灯交叉口自动驾驶车辆精细化路径调度方法。它以自动驾驶车辆为研究对象,对无信号交叉路口进行网格离散化处理,依据路径规划算法获取路径学习数据集。基于该数据集,通过CNN网络构建粗颗粒度的快速路径调度决策模型,对自动驾驶车辆在无信号交叉口的实时路径进行快速决策调度。为实现自动驾驶车辆的精细化路径调度,在粗颗粒度快度路径调度决策模型的基础上,增加了DQN强化学习网络来计算模型决策行为的评价值,构建了奖励函数以及使用自适应学习率来提高模型预测的准确率,平衡自动驾驶车辆在无信号交叉口中路径规划调度的实时性和精准性。本发明能够提升自动驾驶车辆在通过无信号灯交叉路口时的通行效率。
Signal-lamp-free intersection automatic driving vehicle refined path scheduling method based on double learning networks
基于双学习网络的无信号灯交叉口自动驾驶车辆精细化路径调度方法
2023-02-24
Patent
Elektronische Ressource
Chinesisch
Europäisches Patentamt | 2023
|Signal-lamp-free intersection vehicle safe passing method
Europäisches Patentamt | 2020
|Vehicle scheduling method and system for intersection without signal lamp
Europäisches Patentamt | 2024
|Europäisches Patentamt | 2023
|Europäisches Patentamt | 2020
|