基于深度学习的智能车辆辅助驾驶系统设计

Design of Intelligent Vehicle Assistance Driving System Based on Deep Learning

为帮助驾驶员在夜间行车时确认前方路况,设计了保证自己及他人出行安全的车载夜视辅助驾驶系统。通过构建深度学习神经网络算法,将经预处理得到的车辆周围图像作为神经网络的输入数据,经逐层训练与调参得到离线网络运算,从实时图像数据中提取障碍物的特征信息与运动情况。通过驾驶车辆上采集设备探测反馈行车轨迹环境,正确识别车辆所处的环境状态,进而提醒驾驶员,并帮助车辆面对突如其来的危险时采取正确行驶决策,避免事故的发生。

In order to help drivers confirm the road condition in front of them when driving at night,an on-board night vision assistant driving system is designed to ensure the safety of their own and others.By constructing a deep learning neural network algorithm,the pre-processed image around the vehicle is used as the input data of the neural network,and the off-line network operation is obtained through layer-by-layer training and parameter adjustment.The feature information and movement of the obstacle are extracted from the real-time image data.By detecting and feeding back the trajectory environment of the vehicle through the collection equipment on the driving vehicle,we can correctly identify the environmental state of the vehicle,and then remind the driver,and help the vehicle to take correct driving decisions when facing sudden dangers,so as to avoid accidents.