基于机器学习的车辆路面类型识别技术研究
发布时间：2019-03-26 08:53
【摘要】：当车辆在各种不同的路面上行驶时,获知路面类型信息将有助于提高乘车人的安全性和舒适性,不同的路面类型将对车辆的加速、制动及操控等驾驶策略产生影响。基于机器学习的基本原理,提出一种使用加速度传感器和相机特征数据融合对路面类型进行分类的方法,并与单独使用其中一种传感器进行了比较。使用垂直加速度和车速数据并利用车辆动态模型还原路面轮廓,进而完成特征提取和路面类型分类;对相机采集的路面图像数据进行特征提取和分类;将两类传感器的数据特征进行融合,完成路面类型识别任务。实验结果表明:使用两种传感器数据特征融合的方法,不但识别精度有所提高,而且其可靠性和适应性也都优于单独使用加速度数据或路面图像数据。
[Abstract]:When vehicles travel on different roads, knowing the information of road types will help to improve the safety and comfort of passengers, and different road types will affect the driving strategies of vehicles such as acceleration, braking and handling. Based on the basic principle of machine learning, a method for classification of road surface types using accelerometer and camera feature data fusion is proposed and compared with one of the sensors. Using the vertical acceleration and speed data and using the vehicle dynamic model to restore the road contour, then complete the feature extraction and road type classification, and carry on the feature extraction and classification to the road surface image data collected by the camera. The data features of the two types of sensors are fused to complete the road type identification task. The experimental results show that not only the recognition accuracy is improved, but also the reliability and adaptability are better than the acceleration data or road image data by using the two kinds of sensor data feature fusion method.
【作者单位】： 长春理工大学光电工程学院光电工程国家级实验教学示范中心;
【基金】：吉林省自然科学基金项目(20150101047JC)
【分类号】：TP391.41;U463.6
，
本文编号：2447380
[Abstract]:When vehicles travel on different roads, knowing the information of road types will help to improve the safety and comfort of passengers, and different road types will affect the driving strategies of vehicles such as acceleration, braking and handling. Based on the basic principle of machine learning, a method for classification of road surface types using accelerometer and camera feature data fusion is proposed and compared with one of the sensors. Using the vertical acceleration and speed data and using the vehicle dynamic model to restore the road contour, then complete the feature extraction and road type classification, and carry on the feature extraction and classification to the road surface image data collected by the camera. The data features of the two types of sensors are fused to complete the road type identification task. The experimental results show that not only the recognition accuracy is improved, but also the reliability and adaptability are better than the acceleration data or road image data by using the two kinds of sensor data feature fusion method.
【作者单位】： 长春理工大学光电工程学院光电工程国家级实验教学示范中心;
【基金】：吉林省自然科学基金项目(20150101047JC)
【分类号】：TP391.41;U463.6
，
本文编号：2447380
本文链接：https://www.wllwen.com/kejilunwen/ruanjiangongchenglunwen/2447380.html
