非理想通信下多层级车辆群智决策多车控制数据集

本数据集围绕国家重点研发计划项目“智能汽车云控平台架构、基础软硬件及核心应用技术研究”中“云控智能车辆的车路云协同决策、规划与控制”课题，在非理想通信条件下构建了多层级车辆群智决策多车控制的实车与云服务器在环综合数据体系。一方面，基于真实云平台与SUMO联合搭建云服务器在环试验平台，在城市交叉路口场景下开展多工况仿真测试，采集不同交通流量和受限通信条件下车辆运行状态、交叉路口冲突点相对距离轨迹以及云端控制指令与通信日志，实现对“车辆间不发生碰撞、平均车速控制误差≤2km/h”等指标的可量化评价。另一方面，在荣乌高速新线固安南服务区附近开展云控三车队列实车测试，在同一路段、同一三车编队条件下分别采集“正常通信工况”和“受限信息工况（平均时延≥150ms，丢包率＞15%）”两组数据，每组包含多次重复试验，记录队列三车的方向盘转角、车速、纵向加速度、车间距，以及边缘云端下发的推荐速度指令和车云通信报文时间戳，用于精确计算通信时延和丢包率，构造平均车速控制误差与最小安全车间距等评价指标。数据处理环节对实车和云服务器在环数据统一完成时间同步、单位标准化、滤波降噪、异常值识别与缺失片段插值，给出结构化字段说明和工况级统计指标，可为云控自动驾驶多车协同控制算法评估、非理想通信补偿方法对比、多层级群智决策建模与仿真，以及车路云协同控制技术的项目验收与标准制定提供数据支撑。This dataset is developed under the framework of the project "Research on Architecture, Basic Software and Hardware and Core Application Technologies of Intelligent Vehicle Cloud-controlled Platform" under the National Key R&D Program of China, specifically targeting the subject of "Vehicle-Road-Cloud Collaborative Decision-making, Planning and Control for Cloud-controlled Intelligent Vehicles". It constructs a comprehensive real vehicle and cloud server-in-the-loop data system for multi-level vehicle swarm intelligent decision-making and multi-vehicle control under non-ideal communication conditions. On one hand, a cloud server-in-the-loop test platform is jointly built based on a real cloud platform and SUMO. Multi-condition simulation tests are carried out in urban intersection scenarios, collecting vehicle operating status, relative distance trajectories of intersection conflict points, cloud-side control instructions and communication logs under different traffic flows and restricted communication conditions, enabling quantifiable evaluation of indicators such as "no collision between vehicles, average vehicle speed control error ≤ 2 km/h". On the other hand, real vehicle tests of a cloud-controlled three-vehicle platoon are conducted near Gu'an South Service Area on the new line of G18 Rongwu Expressway. Under the same road section and same three-vehicle formation conditions, two sets of data are collected respectively: "normal communication condition" and "restricted information condition (average delay ≥ 150ms, packet loss rate > 15%)". Each set includes multiple repeated tests, recording the steering wheel angle, vehicle speed, longitudinal acceleration, vehicle spacing of the three platoon vehicles, as well as the recommended speed instructions issued by the edge cloud and the timestamp of vehicle-cloud communication messages. These data are used to accurately calculate communication delay and packet loss rate, and construct evaluation indicators such as average vehicle speed control error and minimum safe vehicle spacing. In the data processing stage, unified time synchronization, unit standardization, filtering and noise reduction, outlier identification and missing segment interpolation are performed on the real vehicle and cloud server-in-the-loop data. Structured field descriptions and working condition-level statistical indicators are also provided. This dataset can provide data support for the evaluation of cloud-controlled autonomous driving multi-vehicle collaborative control algorithms, comparison of non-ideal communication compensation methods, modeling and simulation of multi-level swarm intelligent decision-making, as well as project acceptance and standard formulation of vehicle-road-cloud collaborative control technologies.

搜集汇总

背景与挑战

背景概述

该数据集源于智能汽车云控平台相关国家重点研发计划项目，针对非理想通信条件，通过云服务器在环仿真和实车测试采集了多层级车辆群智决策与多车控制数据。数据经过统一处理，可用于协同控制算法评估、通信补偿方法对比及车路云协同技术标准制定。