本研究目的為利用自走車機器人在室內環境中擷取周圍場景資料，以建構自走車機器人目前所處之環境的三維紋理貼圖模型。我們成功整合自走車機器人Pioneer 3-DX、雷射測距儀LMS-200、球形全向攝影機Ladybug2完成包含自動擷取三維建模資料、自動建立三維模型、自動紋理貼圖、自動拼貼多位置模型功能之系統。本研究利用自走車機器人所搭載之球形全向攝影機快速擷取周圍環境之環場全景圖，接著讓自走車機器人自轉360度，同時利用搭載之雷射測距儀快速掃描整個周邊環境以擷取三維點雲資料。經由點雲資料三角網格化之演算法，將場景網格模型繪製於基於OpenGL所設計的三維影像顯示系統，並且透過所提出的對於球形全向攝影機與雷射測距儀之校正方法，將環場全景圖之紋理貼合於場景之網格模型上。在擷取多個建模資料後，我們亦透過所提出的拼貼多位置三維模型演算法，將各位置之三維模型拼貼為單一模型，此方法包含基於法向量群偏轉角度之細部校正演算法，校正各位置模型的偏轉角度；以及另一個所提出的基於點集合偏移距離計算之細部校正演算法，校正各位置模型的偏移距離。此外，我們所提出之拼貼多位置三維模型演算法的時間複雜度為O(nlogn)，而迭代最近點(Iterative Closest Point, ICP)演算法的時間複雜度為O(n^2)，兩者比較後得知，本研究所提出的拼貼多位置三維模型演算法比ICP演算法在效能上更為優異。

The purpose of this research is to construct the three-dimensional model including texture of environment utilizing mobile robot in indoor environment. We have integrated mobile robot (Pioneer 3-DX), laser range finder(LMS-200) and spherical camera(Ladybug2) to develop a complete system supporting automatic capturing three-dimensional model data, automatic 3D model construction, automatic texture mapping, and automatic multi-field 3D model registration functions. In this system, by using spherical camera, the mobile robot can capture panoramic picture of its surrounding environment. After this, the robot rotates 360 degree and simultaneously captures 3D point cloud of the environment by utilizing laser range finder. With the algorithm of triangulating point cloud, the proposed system renders the 3D mesh on OpenGL-based rendering module, and mapping the panoramic image to 3D mesh via the proposed calibration method, which performs calibration between the spherical camera and the laser range finder. After capturing sets of data from few adjacent positions, we developed an algorithm to register multi-field 3D model and to merge 3D models of different positions into a integrated one. This method includes a rotation calibration algorithm based on normalize vectors set, and a translation calibration algorithm based on point set, which also calibrates translation distance. In addition, we compare the multi-field 3D model algorithm with ICP (Iterative Closest Point) algorithm. According to our analysis, the complexity of the proposed algorithm is O(nlogn) and the complexity of ICP algorithm is O(n^2). Thus, the proposed algorithm has higher performance.

[1] “Google Trekker,” 2013, http://www.google.com/help/maps/streetview/mobile/learn/detail/trekker.html#
[2] P. Henry, M. Krainin, E. Herbst, X. Ren, D. Fox, “RGB-D mapping: Using Kinect-style depth cameras for dense 3D modeling of indoor environm- ents, ”The International Journal of Robotics Research, Vol. 31, No. 5, pp. 647-663, 2012.
[3] A. Banno, T. Masuda, T. Oishi, K. Ikeuchi, “Flying Laser Range Sensor for Large-Scale Site-Modeling and Its Applications in Bayon Digital Archival Project, ” International Journal of Computer Vision, Vol. 78, Issues 2-3, pp. 207-222, 2008.
[4] D. Chrysostomou, A. Gasteratos, L. Nalpantidis, G. C. Sirakoulis, “Multi-view 3D scene reconstruction using ant colony optimization techniques, ” Measurement Science & Technology, Vol. 23,No. 11, 2012.
[5] 楊雄壹, ”利用自走車機器人搭載雷射測距儀快速建立三維室內環境模型的方法”, 碩士論文,國立台灣科技大學, 2011.
[6] P. Biber , H. Andreasson , T. Duckett , A. Schilling , “3D modeling of indoor environments by a mobile robot with a laser scanner and panoramic camera, ” in Proceedings of the IEEE/RSJ International Conference on Intelligent Robots and Systems, Vol. 4, pp. 3430-3435, 2004.
[7] H. Andreasson, A. J. Lilienthal, “6D scan registration using depth-interpolated local image features, ” Robotics and Autonomous Systems, Vol. 58, Issue 2, pp. 157-165, 2010.
[8] Y. Bok, Y. Hwang, I. S. Kweon, “Accurate Motion Estimation and High-Precision 3D Reconstruction by Sensor Fusion,” in Proceedings of 2007 IEEE International Conference on Robotics and Automation, Roma, pp. 4721-4726, 2007.
[9] O. Wulf, B. Wagner, “Fast 3D-scanning methods for laser measurement systems, ” in Proceedings of International Conference on Control Systems and Computer Science(CSCS14), pp. 2-5, 2003.
[10] S. Fleck , F. Busch, P. Biber, W. Straser , “Graph cut based panoramic 3D modeling and ground truth comparison with a mobile platform -The Wagele, ” Image and Vision Computing, Vol. 27, Issues 1-2, pp. 141-152, 2009.
[11] M. Sheehan, A.Harrison, P. Newman, “Self-calibration for a 3D laser, ” International Journal of Robotics Research archive, Vol. 31, Issue 5, pp. 675-687, 2012.
[12] P. J. Besl, N. D. Mckay, “A Method for Registration of 3-D Shapes, ” IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 14, No. 2, 1992.
[13] A. E. Johnson, S.B. Kang, “Registration and integration of textured 3D data, ” Image and Vision Computing, Vol. 17, Issue 2, pp. 135-147, 1999.
[14] “Panoramic Vision,” 2013, http://homepages.laas.fr/simon/eden/rover/perception/pano.php
[15] S. Teller, “Automated urban model acquisition: Project rationale and status,” in Image Understanding Workshop, pp. 455-462, 1998.
[16] “Omnidirectional camera” 2013, http://vstone.co.jp/products/sensor_camera/download/omni.pdf
[17] K. H. Tan, H. Hua, N. Ahuja, “Multiview Panoramic Cameras Using Mirror Pyramids,” IEEE Transactions on Pattern Analysis and Machine Intelligence Archive, Vol. 26, Issue 7, pp. 941-946, 2004.
[18] “Point Gray innovation in imaging,” 2013, http://ww2.ptgrey.com/PTGREY_Complete_Catalog.pdf
[19] “Pioneer 3-DX,” 2013, http://www.mobilerobots.com/researchrobots/pioneerp3dx.aspx
[20] “Laser Range Finder LMS200,” 2013, http://www.sick.com.tw/pdf/DIVO5/LMS2xx.pdf
[21] Point Grey Research, “Overview of the Ladybug image stitching process,” Technical Application Note TAN2008010, Revised June 19, 2013.
[22] T. Jost, H. Heinz, “Fast ICP algorithms for shape registration,” Lecture Notes in Computer Science, Vol.2449, pp.91-99, 2002.