基于卷积神经网络的驾驶辅助系统设计

Design of driving assistance system based on convolutional neural network

摘要:

为了减轻驾驶员在行驶过程中的操作负担，进而降低误差判断事件的出现几率，设计一种基于卷积神经网络的驾驶辅助系统。在执行良好的汽车导航架构中，限定Learning Navigation模块与Learning Controller模块的连接位置，再根据辅助驾驶传感器对于行驶画面的采集情况，对车辆巡航能力进行定向控制，抑制监测仪表中辅助波的过渡振动，完成驾驶辅助系统的需求与设计分析。在此基础上，确定辅助激活函数、约束仪表中的行车图像，建立标准化的卷积神经网络。按照驾驶辅助数据的学习结果，对其进行传输处理，进而连接驾驶辅助系统的Job请求，实现系统的顺利运行。利用卷积神经网络平台设计实车实验结果表明，应用驾驶辅助系统后，车辆监测仪表中辅助波振动幅度的最小值处于36-61Hz之间，平均波长偏移量明显减小，驾驶员的行驶操作负担得到有效缓解。

Abstract:

In order to reduce the operating burden of drivers in the process of driving and thus reduce the probability of error judgment events, a driving assistance system based on convolutional neural network is designed. In a well-executed automobile Navigation architecture, the connection position between the Learning Navigation module and the Learning Controller module is defined. Then, according to the collection of driving pictures by the auxiliary driving sensors, the directional control of vehicle cruising ability is carried out to suppress the transition vibration of auxiliary waves in the monitoring instrument, so as to complete the requirements and design analysis of the driving assistance system. On this basis, the auxiliary activation function and the driving image in the instrument are determined, and a standardized convolutional neural network is established. According to the learning results of driving assistance data, it is transmitted and processed, and then the Job request of driving assistance system is connected to realize the smooth operation of the system. The experimental results of real vehicle design using convolutional neural network platform show that the minimum vibration amplitude of auxiliary wave in vehicle monitoring instrument is between 36-61hz after the application of driving assistance system, and the average wavelength offset is significantly reduced, thus effectively relieving the operating burden of the driver.