摘要
Abstract
In allusion to the problem that single sensor cannot fully perceive the information of automotive driving environment,which leads to high missing alarm rate of collision warning,an automotive collision accident warning system based on multi-sensor fusion is designed.The environmental sensing module of the system can collect signal data of the front and side environment information of the automotive by means of millimeter wave radar sensor,camera,and ultrasonic sensor,and send it to the multi-sensor processing module of the data processing unit.In this module,the automotive collision accident risk prediction model based on multi-sensor information decision fusion network is used to extract the multi-sensor features by means of the pre-processing layer.Each type of sensor data is processed independently by the sub-decision layer to predict collision risk.The multi-risk prediction results of the sub decision layer are summarized at the comprehensive decision layer to evaluate the collision accident risk of the automotive.Based on the predicted risk results,whether there is an accident risk is determined and warning messages are sent timely.The experimental results show that under the application of this system,the missing alarm rate of automotive collision accident warning is reduced,which can improve the efficiency of automotive safety traffic.关键词
多传感器/汽车碰撞/事故预警/毫米波雷达传感器/超声波传感器/风险预测/安全阈值/神经网络结构Key words
multi-sensor/automotive collision/accident warning/millimeter wave radar sensor/ultrasonic sensor/risk prediction/safety threshold/neural network architecture分类
电子信息工程
