基于集散决策体系结构的智能车辆自主导航

Autonomous Navigation of Intelligent Vehicle Based on Architecture with Central and Distributed Decisions

智能车辆的体系结构作为智能车辆系统的基础,在构建智能车辆前必须得到合理的设计。为保证智能车辆系统的实时性和智能性,提出了基于集散决策的智能车辆体系结构。该结构由信息感知、规划决策、执行3个基本模块组成,其中规划决策分为低层次的分散决策和高层次的集中决策;分散决策对各种环境信息进行并行处理以得到各局部决策结果,集中决策对各分散决策结果进行综合判断并做出最终决策。按照以上设计思想,对道路环境下的智能车辆体系结构进行了仿真,同时实际构建了智能车辆车道识别及跟踪系统的体系结构。并进行了系统设计及实车试验。仿真结果表明,智能车辆能够根据实际环境信息做出合理决策,顺利完成车道跟踪、车距保持、换道行驶等任务。试验结果表明,在该体系结构控制下的智能车辆系统能够准确、可靠地完成车道识别、车道跟踪及车速保持任务。

Architecture of intelligent vehicle system should be appropriately built before the vehicle system is designed. The architecture based on central and distributed decisions was presented. It composed of three basic modules ： perception, decision - making and implementation. The decision -making module was classified into distributed parts and a central part. The distributed parts processed the environment data and made local decisions independently, and the central part made the final decision using all the distributed decisions. Because the local decisions were made at the same time by several modules, and the final decision - making was simple logical judgment, both the intelligence and real - time processing of the vehicle system could be guaranteed. The mechanism of the architecture for vehicle navigation was simulated. The architecture for lane recognition and tracking was built, and the experiment was implemented. The simulation shows the vehicle system can make proper decisions according to changing environment, and implement lane tracking, safe space keeping and lane changing successfully. The experiment results show the intellggent vehicle can recognize and track the lane reliably and accurately.