基于人车信息融合的大型车辆轨迹识别及其通行道路立体线形优化方法研究

为了科学地改善道路交通事故黑点和提升大型车辆运行安全性，本项目拟从优化道路立体线形的角度对大型车辆运行的安全性与舒适性进行研究，并在人车信息融合的背景下提出考虑大型车辆运行宽容性的道路立体线形优化方法。首先，基于人车信息融合的背景，对“人-车-路”系统多源异构数据进行采集和分析，构建考虑驾驶人生理心理特性的大型车辆运行状态识别算法及大型车辆三维运动轨迹模型；然后，引入卷积神经网络、支持向量机、模糊聚类、插值函数等数理方法，对不同道路条件下的“人-车-路”交互作用机理进行研究，构建道路条件与大型车辆三维运动轨迹的关联模型；最后，以关联模型为基础，采用控制变量的方法并借助模拟驾驶及3D可视化技术，搭建考虑大型车辆运行宽容性的道路立体线形设计指标建议库。研究内容涵盖了大型车辆运行状态识别、大型车辆运行轨迹建模、道路立体线形优化设计等方面，成果将可为改善大型车辆运行安全水平提供理论支撑和标准建议。

In order to scientifically ameliorate the black spots of traffic accidents and improve the safety performance of large vehicle operations, this project, from the perspective of optimizing road three-dimensional alignment, studies the safety and comfort issues of large vehicle operations systematically. And in the context of human-vehicle information fusion, we intend to propose a method of optimizing road three-dimensional alignment that considers the tolerance of large vehicle operations. First of all, based on the background of human-vehicle information fusion, multi-source heterogeneous data of ‘human-vehicle-road’ system are collected and analyzed to create a recognition model of large vehicle running state that considers driver's physiological and psychological characteristics. After that, based on mathematical methods such as convolutional neural network, support vector machine, fuzzy clustering and interpolation function, we study the interaction mechanism of ‘human-vehicle-road’ system under different road conditions, and then construct an association model between road conditions and large vehicle trajectories. Finally, based on the association model between road conditions and large vehicle trajectories, we intend to build a road three-dimensional alignment design indicator suggestion base considering the tolerance of large vehicle operation, using control variable method and simulating driving technology. This research covers the recognition of large vehicle running state, the modeling of large vehicle running trajectories, and the optimal design of road three-dimensional alignment, whose results can provide theoretical supports and standard suggestions for ameliorating the safety level of large vehicle operations.