自动泊车环境感知用超声波传感器建模研究

发布时间：2017-12-28 14:17

本文关键词：自动泊车环境感知用超声波传感器建模研究　出处：《中国集成电路》2016年12期 　论文类型：期刊论文

【摘要】：本文提出了一种基于BP神经网络(Backpropagation Neural Network,BPNN)的方法,以此建立超声波传感器仿真模型用于仿真开发自动泊车环境感知算法。为使所建模型能够模拟超声波传感器系统在检测平板障碍物时得到的输出距离,且使该输出能反映障碍物相对于传感器模型的真实位置,对所采集数据样本进行超声测距误差的来源分析和补偿。BPNN模型基于处理后的数据样本进行了训练测试。最终模型在高级驾驶员辅助系统仿真开发环境(Pre Scan)下进行仿真应用验证,结果表明在传感器发射面与障碍物表面相对夹角±40°范围内BPNN方法可行,所建模型具有仿真应用价值。
[Abstract]:In this paper, a method based on BP neural network (Backpropagation Neural Network (BPNN)) is proposed to establish ultrasonic sensor simulation model and simulate the development of automatic parking environment awareness algorithm. In order to make the model can output distance simulation system of ultrasonic sensor in detecting flat obstacles, and the output can reflect the obstacles to the real position sensor model, the analysis of the source and compensation data collected samples for ultrasonic ranging error. The BPNN model is based on the processed data samples for training tests. The final model is verified by simulation application under the advanced driver assistance system simulation development environment (Pre Scan). The results show that the BPNN method is feasible in the range of the angle between the surface of the sensor and the obstacle surface, which is within the range of + 40 degrees, and the model has the application value of simulation.
【作者单位】：同济大学汽车学院;
【分类号】：U463.6;TP212
【正文快照】： 1引言自动泊车环境感知开发,广泛使用超声波传感器来检测障碍物,主要因为超声波传感器相对于激光雷达、红外测距以及摄像头等有多方面优势[1]。而且,随着计算机技术的高速发展,基于模型的开发方法已成为一种趋势。对于泊车环境感知算法开发,该方法论相对于传统实车实验开发有

【相似文献】