基于深度学习的汽车自动驾驶控制系统测试方法研究  

Research on the test method of automotive autonomous driving control system based on deep learning

作　　者：吴文金 Wu Wenjin(Tongji University,Shanghai 201804,China)

机构地区：[1]同济大学,上海201804

出　　处：《汽车知识》2025年第2期84-86,90,共4页AUTOMOTIVE KNOWLEDGE

摘　　要：深度学习技术的应用显著提升了自动驾驶控制系统在复杂环境中的稳定性和安全性。本文利用卷积神经网络(CNN)和循环神经网络(RNN)对静态图像数据和时间序列数据进行联合分析,构建多维度数据集,在仿真平台上测试系统性能。实验结果显示:该方法在目标检测准确率、响应时间、能源利用率、极端环境下的鲁棒性和安全接管时间等多项指标上均优于传统方法,表明深度学习模型在增强自动驾驶系统适应能力和响应效率方面具有优势。The application of deep learning technology has significantly improved the stability and safety of autonomous driving control system in complex environments.Convolutional Neural Network(CNN)and Recurrent Neural Network(RNN)were used to jointly analyze static image data and time series data,construct multi-dimensional datasets,and test the system performance on the simulation platform.Experimental results show that the proposed method is superior to the traditional method in many indicators such as object detection accuracy,response time,energy utilization,robustness in extreme environment,and safe takeover time,indicat-ing that the deep learning model has advantages in enhancing the adaptability and response efficiency of the autonomous driving system.

关 键 词：深度学习 汽车自动驾驶 控制系统 测试方法 系统可靠性 

分 类 号：U471.1[机械工程—车辆工程]