The invention discloses a semantic reasoning-based queuing vehicle identification method, system and device and a medium, and the method is used for identifying whether an intelligent driving vehicle is a real parking vehicle or a queuing vehicle or a moving vehicle, and through receiving the perception information of the surrounding environment and combining with the lane-level topological relation given by a map, the intelligent driving vehicle is identified to be the real parking vehicle or the queuing vehicle or the moving vehicle. Queuing labels are analyzed through reasoning semantics of vehicles, more environment information, map topology information and common knowledge of people in intelligent driving are considered, training of a large amount of data is not needed, the requirement for computing power is lower, meanwhile, the identification accuracy of the queuing labels is higher, the method is suitable for different scenes, robustness to fusion noise is higher, and in addition, the method is suitable for popularization and application. By adopting the cooperation of the queuing label judging device and the queuing label arbiter, manual marking is not needed, the ambiguity condition in the same area does not exist, and a safer and more efficient driving strategy can be decided.
本发明公开一种基于语义推理的排队车辆识别方法、系统、设备及介质,该方法用于识别智能驾驶车辆为真停车辆或排队车辆或运动车辆,通过接收周围环境的感知信息,再结合地图给出的车道级拓扑关系,通过对车辆所处语义进行推理分析排队标签,考虑更多环境信息、地图拓扑信息和人在智能驾驶中的常识,无需大量数据进行训练,且算力要求更低,同时排队标签的识别正确率更高,适应于不同的场景,对融合噪声的鲁棒性更强,另外,采用排队标签判定器和排队标签仲裁器的配合,无需人工标记,不存在相同区域出现二义性的情况,有利于决策出更加安全高效的行车策略。
Queued vehicle identification method, system and equipment based on semantic reasoning and medium
基于语义推理的排队车辆识别方法、系统、设备及介质
2023-10-20
Patent
Elektronische Ressource
Chinesisch