针对当前“请求-响应”被动式行车服务存在的个性化、人性化、实时性等不足,提出了基于多智能体(multi-agent system,MAS)分层集中控制的“自动识别-主动推送”形式的行车主动服务系统(driving active service system,DASS).DASS将行车环境实体与服务作为建模对象,提出基于规划图的行车服务需求模型及用户需求与服务能力匹配算法.以交通安全预警和实时动态路径规划为例,通过自主搭建的驾驶员在环半实物联合仿真平台对DASS典型应用展开实验研究,充分验证了DASS的实时性、主动性与高效性.这对于降低交通拥堵、提升通行效率,加强车辆安全以及改善驾驶体验具有重要价值. In view of the existing personalization, humanization and real-time of the current “request-response” passive driving service, a multi-agent system (MAS) hierarchical automatic control- Driving-type active service system (DASS) .DASS takes the driving environment entity and service as the modeling object, and proposes the driving demand model based on the plan and the matching algorithm of the user demand and service ability.According to the traffic Safety early warning and real-time dynamic path planning as an example, DASS’s real-time performance, initiative and efficiency are fully verified through the self-built driver-in-the-loop simulation platform for DASS, which can reduce the traffic jam , Improve traffic efficiency, enhance vehicle safety and improve the driving experience is of great value.