改良式的多分類器路面辨識應用於遠端呼叫自走車的導航研究

Study on an Improved Multi-Classifier Road Detection Applied to Navigation of Automatic Land Vehicle through Remote Calling

沈思瑜(Szu-Yu Shen)

指導教授 ： 駱榮欽

國立臺北科技大學/電資學院/電腦與通訊研究所/碩士(2011年)

https://doi.org/10.6841/NTUT.2011.00573

摘要

本研究提出一種能遠端呼叫自動導航車(ALV)的架構。首先，呼叫端經由無線網路傳送目的地之GPS訊息予ALV，藉由已知的地圖資訊，以迪杰斯特拉演算法進行路徑規劃。行進中所需要的路面資訊，使用AdaBoost演算法來進行路面辨識，對不同的路面種類進行訓練和分類，可找到可行進的路面區域。藉著路面辨識的結果，ALV可決定一個行進方向。由路徑的資訊取得目標子節點的位置以及自動車目前位置的差異，可找到該行進的絕對方向，而與電子羅盤偵測到的車輛行進方向比較，可找到車輛該偏轉的方向。而往目標前進時，以繞道閃避的方式避開AdaBoost找出的非路與立體障礙物。藉由此流程，自動車可到達呼叫端的位置。

關鍵字

電腦視覺 ； 影像處理 ； 自動導航車 ； AdaBoost ； 多分類器

並列摘要

This research proposes a technique of calling an ALV remotely. Firstly, the calling side sends the GPS information of the target to the ALV through wireless network, then path planning is completed by prepared map data with Dijkstra algorithm. AdaBoost algorithm is introduced for implementing road detection; different types of classifiers are trained to detect different road surfaces. With the result of road detection, the ALV is able to find the drivable region eliminating the obstacles and not-road region. By determining the difference with the position of the ALV and current sub-goal on the planned path, the forward direction to the sub-goal is decided; with comparing it to the driving direction detect by digital compass, the direction of deflection can be decided. When ALV drives to the sub-goal, obstacles and not-road region is avoided by surpass. In conclusion, ALV can reach the calling side.

並列關鍵字

Computer Vision ； Image Processing ； Automatic Land Vehicle ； AdaBoost ； Multi-Classifier

參考文獻

[2] Hui Kong, Jean-Yves Audibert, and Jean Ponce, “General Road Detection From a Single Image,” in IEEE Transactions on Image Processing, 2008 pp, 2211-2220.

[3] Yaniv Alon, Andras Ferencz and Amnon Shashua, “Off-road Path Following using Region Classification and Geometric Projection Constraints,” in IEEE Computer Society Conference on Computer Vision and Pattern Recognition, vol. 1, 2006, pp. 689-696.

[4] Sung-Soo Kim and Jong-Hyun Lee, “A study on design of dynamic route guidance system using forecasted travel time based on GPS data and modified shortest path algorithm,” in Proc. IEEE/IEEJ/JSAI International Conference on Intelligent Transportation Systems, 1999, pp. 44-48.

[8] Yoav Freund and Robert E. Schapire, “Experiments with a New Boosting Algorithm,” in Proc. 13th Int'l Conf. Machine Learning, 1996, pp. 148-156.

[9] Ming-Kuei Hu, “Visual Pattern Recognition by Moment Invariants,” IRE Transactions on Information Theory, 1962, pp. 179-187.