《电子技术应用》
您所在的位置:首页 > 人工智能 > 设计应用 > 低成本智能小车控制感知预测多路径通信方法
低成本智能小车控制感知预测多路径通信方法
集成电路应用
段皓淇,郭磊
新乡工程学院
摘要: 为解决低成本智能小车本地控制与高动态通信难以协同、路径频繁切换的问题,该方法构建了车载控制、外接通信和路侧边缘计算三层架构,提出控制感知风险最小化预测多路径的方法(CARMPM);综合路径当前与预测可靠性、寿命、时延、跳数、业务类型和切换代价评分,并采用业务分级滞回。结果表明:在3种车辆密度、4种策略和15个固定随机种子的轻量级仿真中,与反应式策略相比,路径切换次数降低76.9%~89.2%,投递率差值为-0.55~0.01个百分点,平均时延增加4.16~4.57 ms。该方法可在基本保持通信可靠性的同时抑制路径抖动,为低成本智能小车的车路协同与硬件在环验证提供分层方案。需要明确的是,该文中的绝对投递率是抽象链路模型的内部指标,不代表实际道路VANET性能。
中图分类号:TP393.02文献标识码:ADOI:10.19339/j.issn.1674-2583.2026.09.007
中文引用格式:段皓淇,郭磊. 低成本智能小车控制感知预测多路径通信方法[J].集成电路应用,2026,43(9):45-50.
英文引用格式:Duan Haoqi,Guo Lei. Control-aware predictive multipath communication method for low-cost intelligent vehicles[J].Application of IC,2026,43(9):45-50.
Control-aware predictive multipath communication method for low-cost intelligent vehicles
Duan Haoqi,Guo Lei
Xinxiang Institute of Engineering
Abstract: To address the difficulty of coordinating the local control and highly dynamic communication of low-cost intelligent vehicles together with the problem of frequent route switching, a three-layer architecture comprising onboard control, external communication, and roadside edge computing was constructed, and a controlaware risk-minimizing predictive multipath method (CARM-PM) was proposed. The method combines current and predicted path reliability, lifetime, delay, hop count, service type, and switching cost, and applies service-specific hysteresis. In lightweight simulations with three vehicle densities, four routing strategies, and fifteen fixed random seeds, CARM-PM reduced path switches by 76.9%~89.2% compared with reactive routing. The packet delivery ratio changed by -0.55 to +0.01 percentage points, while mean delay increased by 4.16~4.57 ms. The method suppresses route oscillation while maintaining broadly comparable communication reliability, providing a layered approach for vehicle-infrastructure cooperation and hardware-in-the-loop validation of lowcost intelligent vehicles.
Key words : intelligent vehicle; vehicular ad hoc network; predictive multipath routing; MSP430F5529; edge computing; control awareness

引言

低成本智能小车通常由红外传感器、超声模块和编码器完成环境感知,再由微控制器执行循迹、测距、测速与电机驱动。小车接入车载自组网(Vehicular Ad Hoc Network,VANET)后,系统同时面临两类时序约束:本地控制必须以毫秒级周期稳定运行,协同通信则受到节点移动、链路断续和多跳竞争影响。位置路由无需维护完整拓扑,多路径机制还能在链路失效前准备替代路径,适合高动态环境[1-2];但若选路仅追求瞬时低时延或最长寿命,频繁改写转发表会引入控制消息、缓存失配和业务抖动[3-5]。

Saleh(2023)[1]将车辆跟驰预测用于服务质量感知的多路径切换,Li(2020)[6]给出了基于MSP430F5529的智能小车硬件。前者未讨论资源受限控制器承受的路由切换负担,后者未配置原生车联网通信,二者之间仍缺少控制闭环与预测选路的可实现连接。为此,本文提出控制感知风险最小化的预测多路径方法(ControlAware RiskMinimizing Predictive Multipath,CARMPM):将感知控制、无线接入和边缘选路分层;把业务类型与切换代价写入路径评分;在统一随机输入下比较四种策略。研究目标不是使单一指标最优,而是在可靠性、时延与路径稳定性之间获得可解释折中。


本文详细内容请下载:

https://www.chinaaet.com/resource/share/2000007330


投稿邮箱:ICAPP@belling.com.cn


作者信息:

段皓淇,郭磊

(新乡工程学院,河南新乡453000)

2.jpg

此内容为AET网站原创,未经授权禁止转载。