演化式物體影像追蹤與傾斜定位

Evolution-based object tracking and localization based on tilt photographing

鄭明育(Ming-Yu Cheng)

指導教授 ： 許陳鑑

淡江大學/工學院/電機工程學系碩士班/碩士(2009年)

https://doi.org/10.6846/TKU.2009.00183

摘要

機器人的定位(Localization)，其用意在於得知障礙或目標的座標資訊，進而規劃路徑達到避障以及追蹤的效果，本論文為了摒除多感測器在資料整合上花費較多時間處理的缺點，以及提升機器人對於擬人化的發展與需求，將以CCD為主要的感測器在追蹤演算法及傾斜定位上提供不同以往的研究內容。其中追蹤演化法的研究中，考量到粒子濾波器(PF)所得到的結果會因環境因素而無法進行追蹤，因此本文提出了一種以粒子群聚最佳化(PSO)為搜尋主體並結合單體演算法(NM)的複合式(Hybrid)追蹤演算法則，在此法則下本文除了克服NM於影像中容易落於區域最佳解的缺陷，並成功的以此複合式演算法增加了處理資料量以及追蹤成功率。同時文中於傾斜定位上詳細描述藉由已知CCD架設的高度以及推算出各像素的照射角度，來完成已知目標物影像座標下的空間定位以及量測原理，並可結合地標(Landmark)完成傾斜角的自我量測，使機器人能達到多種傾斜角的定位。

關鍵字

目標追蹤 ； 追蹤演算法 ； 影像式量測 ； CCD 攝影機 ； 定位

並列摘要

This thesis investigates two important issues of robot navigation: object tracking and localization. In the object tracking system, we propose a hybrid evolutionary algorithm incorporating NM simplex method and particle swarm optimization to improve tracking performance in terms of processing speed and successful rate in comparison to particle filter (PF) tracking for complex environment. As for the problem of robot localization, this thesis provides an image-based localization method based on tilt photographing of a single CCD camera. Image captured by the CCD camera is pre-processing to locate the target object in the picture in terms of pixel count deviation from the CCD camera. By using an established formula based on relationship between tilt angle of the CCD camera and distance, coordinate of the target object can be calculated.

並列關鍵字

Object Tracking ； evolution algorithm ； Image-based Measurement ； CCD camera ； localization

參考文獻

[1] C.C. Tsai and C.J. Wu, “Localization of an Autonomous Mobile Robot Based on Ultrasonic Sensory Information,” Journal of Intelligent and Robotic Systems, Vol. 30, No.3, pp.267-277, 2001.

[2] H.H. Lin, Ching-Chih Tasi, Jui-Cheng Hsu and Chih-Fu Chang, “Ultrasonic self-localization and pose tracking of an autonomous mobile robot via fuzzy adaptive extended information filtering,” IEEE International Conference on Robotics and Automation, Taipei, Taiwan, 2003, Vol. 1, pp.1283-1290.

[3] P. Krammer and H. Schweinzer, “Localization of object edges in arbitrary spatial positions based on ultrasonic data,” IEEE Sensors Journal, Vol. 6, No. 1, pp.203-210, 2006.

[4] J. Canou, C. Novales, G. Poisson, and P. Marche, “Quick primitives extraction using inertia matrix on measures issue from an ultrasonic network,” IEEE International Conference on Robotics and Automation, Seoul, Korea, 2001, Vol. 4, pp.3999-4004.

[5] E. Menegatti, A. Pretto, A. Scarpa and E. Pagello, “Omnidirectional vision scan matching for robot localization in dynamic environments,” IEEE Transactions on Robotics, 2006, Vol. 22, no. 3, pp. 523-535.