车路协同下多智能车辆大范围渐进稳定模态的聚合解析与自适应控制

Multi-intelligent vehicles' safe driving and cluster control, which is a class of autonomous mode for multi-agent sets in car networking environment. It can carry out the resource's self-demand collaboration flexibly in open and dynamic conditions, and improve the transportation system interactivity, security and intelligence effectively...For describing the nonholonomic constraint process of mixed intelligent vehicles fleet, we use differential game theory analysis the behavior and decision-making basis of autonomous body. And we study the inconsistencies between local goals and common goals during collaborative process, to resolve resources effective polymerization problem in the Cooperative Vehicle Infrastructure System（CVIS）.Based on modern control theory and dynamic continuous space-time information, we establish non-linear mathematical model for the self-driving hybrid fleet. It is useful for exploring multi-modes modal evolution laws in this autonomous system with global stability, and the uniform convergence of stable modals near equilibrium point is validated by through collaborative neural network rules. Finally, The model reference adaptive controller was designed which focusing on real-time dynamic operating conditions, then we use simulation and experiment tools verify the robust stability and dynamic performance of co-driving hybrid controller for mixed intelligent vehicles fleet. ..Based on above research achievements, the result of this project can provide theoretical support for multi-intelligent vehicles and collaborative mixed fleet in the CVIS. Besides, it is very useful and practical meaningful to improve the active safety of driving vehicle, so the traffic accident can be reduced and prevented effective.

多智能车辆的安全行驶与集群控制，是一类车联网架构中多智能体元素集的自治模式，能实现开放动态条件下资源有序、灵活、按需的自主协作，以有效提高交通运输体系的交互性、安全性、智能性。.针对智能车辆混合编队协同控制的非完整约束描述过程，运用主从微分对策理论剖析自治主体的行为决策依据，研究其协同进程中局部目标与共同目标的不一致性问题，进而解析车路协同环境下资源的有效聚合；基于现代控制理论和动态连续时空信息，建立自治行驶混合编队的非线性数学模型，探索该类自治系统大范围渐进稳定模态间的多模演化规律，并通过协同神经网络规则验证平衡点稳定模态的一致收敛性；最后，围绕实时工况设计动态寻优的模型参考自适应控制器，采用仿真实验手段验证自主协同混合编队行驶控制器的鲁棒稳定性和动态性能。.拟通过上述研究，为实现基于车路协同的多智能车辆混合编队行驶控制提供理论支撑，对提高车辆行驶中事故主动预防能力具有重要的现实意义。

近年来，由于公路交通事故呈不断上升趋势，已经成为社会关注的一大公害。多智能车辆的安全行驶与集群控制，是一类车联网架构中多智能体元素集的自治模式，能实现开放动态条件下资源有序、灵活、按需的自主协作，以有效提高交通运输体系的交互性、安全性、智能性。本项目重点解析车路协同下多智能车辆混合编队的资源聚合和非完整约束描述过程，构建智能车路系统中多车混合编队模拟仿真平台，运用数字化仿真手段剖析主从运动车辆的行为决策依据，进而研究多目标下自治车辆行驶混合编队的非线性数学建模、稳定控制、多模演化和控制器性能分析。本课题重点开展的核心研究内容包括：混合交通中无人驾驶车辆汇入车队安全性评估与控制方法研究、面向智能车辆的一种基于激光与视频信息时空数据融合的行人检测方法、超车安全预警分析与纵向安全距离模型的研究、基于混合高斯分布的运动车辆实时检测与跟踪算法、面向车路协同系统的智能网关通信技术研究。. 在项目执行期内，项目负责人和项目组取得了优秀的研究成果，共计发表SCI期刊检索论文5篇、EI 期刊检索论文8篇，获得省部级科技进步奖2项，申请国家发明专利3项，获批成立了省部级重点实验室1个，直接/间接培养硕士研究生、博士研究生和博士后9名。课题立足于通过基础理论分析和仿真试验验证的研究工作，服务和应用于未来车联网下的智能交通安全和车路协同控制领域。拟通过系统化研究，为开发智能车路系统、实现基于车路协同的多智能车辆混合自治编队行驶控制提供理论支撑和学科参考，研究成果对“多移动机器人的柔性编队控制”、“多智能体光机电一体化物理系统的数学描述”、“现代实验交通工程学”也具有一定的借鉴价值。