簡易檢索 / 詳目顯示
| 研究生： |
劉昌倫 Liu, Chang-Lun |
|---|---|
| 論文名稱： |
深度神經網路模型在自動駕駛中的應用 Application of Deep Neural Network Models in Autonomous Driving |
| 指導教授： |
劉晉良
Liu, Jinn-Liang |
| 口試委員: |
陳人豪
Chen, Jen-Hao 陳仁純 Chen, Ren-Chuen |
| 學位類別： |
碩士 Master |
| 系所名稱： |
理學院 - 計算與建模科學研究所 Institute of Computational and Modeling Science |
| 論文出版年： | 2019 |
| 畢業學年度： | 107 |
| 語文別： | 中文 |
| 論文頁數： | 34 |
| 中文關鍵詞： | 自動駕駛 、直接感知方法 、卷積神經網路 |
| 外文關鍵詞： | GoogLeNet, AlexNet, autonomous driving algorithm, fine tune |
| 相關次數： | 點閱：2 下載：0 |
| 分享至: |
| 查詢本校圖書館目錄 查詢臺灣博碩士論文知識加值系統 勘誤回報 |
自動駕駛技術在現今社會中扮演重要的角色，在本論文中，我們使用直接感知方法來執行自動駕駛，以前此領域的文獻專注於卷積神經網路中道路標線和偵測其他車輛的特徵提取，而比較沒有專注於現實情形下的性能以及安全性，我們在此篇論文更改了道路參數的設置，也使用規模更深、更廣的卷積神經網路來建構自動駕駛演算法。
首先，我們選擇 AlexNet 和 GoogLeNet 作為提取圖像特徵的卷積神經網路並且調整了神經網路的架構，再來，設置主要結構為五個道路參數的控制器，並加上了符合現實情形的模擬測距儀器，實驗結果顯示，經過微調之後的 GoogLeNet 模型加上改良的控制器，會優於以前文獻直接感知方法的自動駕駛演算法。
Autonomous driving technology plays an important role in our society. In this thesis, we use direct perception approach to achieve autonomous driving. Previous thesis focused on road markings in convolutional neural network and detecting other vehicle instead of focusing on performance and security in real-world situations. We have changed the setting of road parameters and used a deeper and wider convolutional neural network to construct our autonomous driving system.
First, we select AlexNet and GoogLeNet as the convolutional neural network to extract image features and adjust the structure of convolutional neural network. Then, we set up the driving controller with five road parameters and add the measuring distance instrument from TORCS. The experimental results show that the fine-tuned GoogLeNet model with our driving controller is superior to the autonomous driving algorithm in previous thesis.
[1] Y. James, 卷積神經網路介紹
available at https://medium.com/jameslearningnote/%E8%B3%87%E6%96%99%E5
%88%86%E6%9E%90-%E6%A9%9F%E5%99%A8%E5%AD%B8%E7%BF%92-%E7%AC%AC5-1%E8%AC%9B-%E5%8D%B7%E7%A9%8D%E7%A5%9E%E7
%B6%93%E7%B6%B2%E7%B5%A1%E4%BB%8B%E7%B4%B9-convolutional-neural-network-4f8249d65d4f
[2] C. H. Tseng, 初探卷積神經網路
available at http://chtseng.wordpress.com/2017/09/12/%E5%88%9D%E6%8E%A2%
E5%8D%B7%E7%A9%8D%E7%A5%9E%E7%B6%93%E7%B6%B2%E8%B7%AF
[3] Alisonyang, AlexNet之術式解析
available at https://ithelp.ithome.com.tw/articles/10205088
[4] 趙欣, CNN模型之AlexNet
available at https://zhuanlan.zhihu.com/p/31717727
[5] C. Chen, A. Seff, A. Kornhauser, & J. Xiao, (2015). Learning Affordance for Direct Perception in Autonomous Driving, ICCV2015.
[6] A. Krizhevsky, I. Sutskever, & G. E. Hinton, ImageNet Classification with Deep Convolutional Neural Networks, Adv. Neural Inf. Process. Syst., pp. 1–9, 2012.
[7] M. A-Qizwini, et al., Deep Learning Algorithm for Autonomous Driving using GoogLeNet, IEEE intelligent Vehicles Symposium (IV), 89-96, 2017.
[8] Wikipedia